🎥｜animation

[static niche]

i have a 4090 24g card i can try, do you have a link to a workflow or was that in a1111?

that must be upscaled after the fact, seems too good quality to be 512x512 as a base size

[serene jay]

you can use an sdxl workflow as init image to then feed to the animateDiff chain, but the motion module only works with 1.5 models so far. I know there are various groups working on motion sdxl but that will probably be like hotshotXL, 512 finetuned.

you can get pretty decent outputs out of 1.5 models and if you have a 4090 you can get it up to 768px range and run 32 frames without issues

[static niche]

gotcha

[toxic glen]

Starting to get the hang of this Animatediff stuff

[toxic glen]

Wait so you can provide an init image with Animatediff? In the same way you can with Deforum, basically animating (from) the image?

I've asked about that before and I was told No, so I just assumed you couldn't

[serene jay]

Of course you can! Multiple ways, you can convert to latent and use it straight to the sampler You can use IPAdapters. You can use clip vision and more…

[toxic glen]

Cool I'd like to figure that out then! If you can start from an init image you can flow seemelessly from video into your animation no? That would be my main use for it

[wintry vortex]

Made some funky and colorful visuals that were played behind a DJ booth! If anyone has any tips to make the animation more high quality lmk !!
https://youtu.be/lCfyfp8s9co?si=n0FTRq0vnM9cAE7f

[toxic glen]

https://www.tiktok.com/t/ZT8k4pYxo/

in honor of spookytime

[neat coral]

https://www.youtube.com/watch?v=8DrBFEIjAoM live stream on AnimateDiff

[static niche]

512 vs 768

[serene jay]

Experiment with everything!> Higher resolutions, more steps, different samplers, better models. https://youtu.be/bWsMgAEW7vQ?feature=shared

[void girder]

https://www.youtube.com/watch?v=-OkjBB1WJTY

[stable hawk]

Is this using the audioScheduler nodes?

[manic knot]

Whenever I make an image, and then turn on AnimateDiff the image becomes something completely different and animated. Is this normal? 🤔

[mental pecan]

some stuff I did for burning man this year.. https://youtu.be/mgEDoenFJDw?si=bksOSgyMjrcO_nz8

[round jay]

no, just a overlay and cut, and arranged

[toxic glen]

Started trying to animate with my XL LoRA tonight with Animatediff

[exotic knot]

has anyone used video to video nodes?

[noble bison]

https://youtu.be/cePM-BQR7ys?si=tYxdlMROv_o0h6Hx

[stable hawk]

New Workflow for ComfyUI. Simple Vid2Vid upscaler with film interpolation, Upscale, interpolate, change frame rate, fairly simple to use and resource light. https://civitai.com/models/179731?modelVersionId=201697

[stable hawk]

vid2vid lol

vid2vid

[shrewd crown]

https://youtu.be/RDEERxhY8e0 SD, Defomation

[void girder]

https://www.youtube.com/watch?v=0N6Sz2JXtls

[toxic glen]

animating with my LoRA some more

[gloomy fractal]

Can someone please tell me why this simple workflow in comfyui with animatediff not working?

[toxic glen]

I'd guess that checkpoint doesn't have a VAE baked in

[gloomy fractal]

@toxic glen thank you, but still noluck with vae loaded

if I unplug the animatediff model this setup works as expected:

[shrewd crown]

https://www.youtube.com/watch?v=Nbdmjl5itfs

[gusty seal]

https://www.tiktok.com/@remixroomai/video/7295297998622362912

[rich mirage]

increase batch size to 8+

anyone got some nice img2vid with animatediff prompttravel workflows?

[stable hawk]

https://civitai.com/models/162163/geekyghost-blended-int

This one works well. It can use a single image, but a blended int seems to work best. You can use a reverse of the original image if needed.

[rich mirage]

thanks ill try it out

[frigid tinsel]

HotshotXL + AD using audio wave prompt exploration -upscaled and audio enhanced. Model used - LoveXL+ bunch of loras

[round jay]

Chroma V3 https://huggingface.co/zzzAI19/ChromaFT/tree/main

[tulip storm]

animatediff....so...with some checkpoint i get results that i like......the ONLY change I do...is to check the "enable animatediff" and i get this weird stuff.....i wonder what I am doing wrong here....

i use the term "with some checkpoint" bacause it seems to be happening no matter what checkpoin i use

so in short: if i disable animatediff the images are just fine, whenever i check "enable animatediff" and rerun....it is all weird......i must be something fundamentaly wrong.....but i cant see what lol

the pos and neg prompt is unchanged

1.5

[shy oar]

Waterfalls

[frigid tinsel]

Another experimental HotshotAD run .. maybe needs a different audio wave to push the animation differently and rerun.. (100 frames too less )

[stray tree]

"Chrome Visions" - Hybrid Video
Full 4k: https://www.youtube.com/watch?v=44hId_KYPjY

[azure sky]

Love this. Amazing work mate. Right up my street.

[wooden yacht]

could someone teach me how to make the frames more consistent?

[simple jasper]

Stunning !

[simple jasper]

I'm not at all an expert, but maybe by increasing the cadence, also tweaking your ControlNets and adding more can maybe make the frames more smooths and consistent.
Maybe play more with your prompt, your noises etc... Tons of factors can change the consistency i guess

[wooden yacht]

you're right. ill try that

[serene jay]

interpolation. rife and such 2x to 4x

[wooden yacht]

cursed 2nd half

[simple jasper]

Yea the character is way more consistent !

Especially the boobs part haaha

What did you changed on that one ?

Maybe try to remove the background from your video input for the controlnet and put greenscreen instead, so you can composite it on a generated image background

[blissful haven]

Made a few animations using SD and ComfyUI, could anyone maybe help as to why Discord will not embed them for preview and will only show as a download file. Files have been between 3MB-8MB so I don't think it is due to being to large of a file. TIA!

[foggy imp]

I have made a playlist of animation tutorials for Blender & VRoid, etc, here: https://www.youtube.com/playlist?list=PLbJAuQIEql3YuacjyEedREPnwgljH7CST

[foggy imp]

wip

[foggy imp]

https://youtu.be/u5qnK2iubtk?si=vJfNTS7renU-LMQA

[noble bison]

https://youtu.be/0E7j4c0SvC8?si=J1ElafbtAJaYzMs0

[serene jay]

Latent Nightmares for Helloween https://youtu.be/sUgrJDK5UaY

[simple jasper]

So crazy! What tools are you using to get that result ? 👀

[serene jay]

ComfyUI, AnimateDiff, Controlnets, IP Adapters and a little bit of post

[frozen gate]

first attempt at animating my music

https://www.youtube.com/watch?v=bzYfBderLMo

[fallen girder]

"NeoTokyo 2033" | AnimateDiff Text2video
https://youtu.be/CfxfrqzVQyQ

[main island]

trying to do some animated diff stuff too but i feel with 8gb vram i have no chance

[knotty dust]

https://youtube.com/shorts/F8vXXlUhMKY

[kind basalt]

I've run into this problem with specific checkpoints... Only solution I have found is to change the checkpoint

[kind basalt]

lmfao so trueee

[pastel imp]

can anyone recommend a colab notebook for running animatediff? an auto1111 notebook that'll run it would be cool. or stand alone. please don't make me learn comfyui...

[frigid tinsel]

Funky cool -AD- with HotshotXL prompt manipulation _audio inputs- Checkpoint Model is Gleipnir V104 , WowifierXL LoRa, hotshotxl_mm_v1.pth model in AD, 100 frames interpolated and audio reworked. AD params with motionsccale-1.2 and motion pe stretch at 2;

[shrewd crown]

https://youtube.com/shorts/y6EWMGkFEu8?feature=share

[blissful haven]

New to animations, can you use SDXL with AnimateDiff? Or is it only 512x512 based?

[static niche]

no sdxl directly yet, 768 with 1.5

[blissful haven]

Copy ty for the info!

[noble bison]

https://youtu.be/FUkIhBT2SP8?si=UVX5E5Vzn9rZG-iL

[misty mica]

Hi guys, I will outpaint. Which inpaint model do you recommend?

[vital yacht]

On October 31, 1979, several units of the American military cordoned off a vast territory in Maricopa County, east of the city of Phoenix, ordering the evacuation of the area's inhabitants.

That day 13 people died under mysterious circumstances. The bodies were never returned to their families and the entire area was closed for 3 years, guarded day and night by the military. The authorities officially declared that it was a tragic fatality, caused by an (unspecified) harmful natural gas leaking from underground.

Those who lived in that area are still waiting for the truth today.

https://youtu.be/AC__JUNOhGQ?si=VYqwRzIewQ5Dy4EG

[rich gale]

Since the Diffuse Together challenge I’ve been refining my work— and wanted to share the latest music video https://youtu.be/uGiAAaEGAyw?feature=shared

[fallen girder]

Deforum animation, prev image flow RAFT, sdxl
https://youtu.be/bjhLQ5I7pnE

[noble bison]

https://youtu.be/BGGVpp-nCBg?si=S-IgcGr7z_SX8ng7

[north tangle]

https://youtu.be/WheXYfTI8Xo?feature=shared

[neat coral]

where can I find this node "BatchPromptSchedule"?

[cunning temple]

https://youtube.com/shorts/tacS2STuLP8?si=w6Ycp6Q4BzZzETbQ

[serene jay]

great! glad you're back!! let's keep pushing : )

[shrewd crown]

https://www.youtube.com/shorts/-NhWU52rAZo

[neat coral]

where can I find this node "BatchPromptSchedule"?

[quasi spire]

I made an abomination. I hope you like it https://youtu.be/6MD8pQwbA6M

[serene jay]

https://youtu.be/z9Z0_yyElE4

[neat coral]

https://www.youtube.com/watch?v=Zqb1sFKRBzA

[shrewd crown]

Stable Diffusion IMG2IMG https://youtube.com/shorts/CVh2mIagvIo?feature=share

[trail marsh]

Hey everyone. I was wondering if someone could point me in the right direction to understanding a Lora's influence on the outputs in Animated diff with Automatic 1111. Some give tons of motion and have predictable image outputs while others have opposite effect.

[serene jay]

Maybe someone with better technical knowledge can answer this better but from what I understand, a LoRA shouldn't alter in anyway the animation/motion side, LoRAs happen before going to the motion module and sampler (at the CLIP space), what could happen is that some LoRAs are trained on more strict/fewer parameters and all outputs end up looking mostly the same, but they should produce predictable outputs at the motion side. What does the motion is the motion model and motion loras.

[trail marsh]

THanks for the reply. I just got a reply on another chat where he says resolution may be tied to motion. Sometimes I'll feed a LORA in at 0.1 weight and get stuff flying all over the place. Others at weight of 1 don't end up lookiing like the sample image using the same prompt.

[serene jay]

oh yes LoRAs and motion models are trained at different resolutions, that's something to consider. From my personal experience with AnimateDiff, I can tell you that I can reproduce similar motion from different LoRAs over comfyUI

but mostly on the SD 1.5 side

[fallen girder]

Deforum-to-AnimateDiff

[flat wharf]

Please Subscribe to the chicken! Made it with animateDiff and ComfyUI, enjoy! https://youtu.be/aCqBUMgwNv8?si=pX2JruBhjinMRc0u

[serene jay]

great storytelling, the sound and edit very nice too, congrats!

[kind basalt]

Hi everyone! I have a question... What hardware do you use? I tried to run an animatediff workflow on a local comfyui install and well... Had to turn off the computer. I have 16gb of gpu but it is not Nvidia.

[flat wharf]

Thank you for your feedback 🙂

[flat wharf]

animateDiff is VRAM intensive, it's quite an heavy process even with my 4090 and its 24gb

[kind basalt]

So far I've been experimenting on colab and results and speed are good but the setup and connection instability can be very frustrating at times, plus it's expensive 🫰

[serene jay]

Yeah as Nuke was saying, it's memory hungry like crazy, I had a 3080Ti 12 GB and couldn't do much with that, got me a 4090 and things are much better but still I get out of memory errors!

I'd suggest using low resolutions pixel-wise, don't go for 768 pixels or higher, use resolutions closer to 512px and then upscale

[kind basalt]

Yeah it's what I tried hahah it was not a success, my gpu is from 2017 and not an Nvidia....

I guess I will wait for my next refill of colab compute and come up with cool ideas until then 😆

[serene jay]

also, if you have memory issues, try running fewer frames

you can then interpolate with RIFT or something like that

with 2X you can render let's say 8 frames and then interpolate 16 from those

I don't know what animateDIff nodes are you using but on the advanced ones you have "uniform context options" and you can provide a context lenght, let's say your card maxes at 12 frames

you put that lenght there and it will run the sampler every 12 frames and then concatenate

the combiner will produce the lenght you want automatically

48-64 frames or more

it'll be slow but you can run more frames

[rustic trail]

eggcellent!

[kind basalt]

Here's my first "thing" i put out 🙂

https://www.instagram.com/reel/CzbN14Qoc8B/?igshid=MWs3ZDFiZHIydnlsOQ==

datariff_ai

Ideas

Mi primera publicación en este proyecto, enfocado a la creación artística con Inteligencia Artificial. Después de varios meses explorando y experimentando me lanzo a la piscina, espero que os guste!

Creado con #Animatediff en #comfyUI y la música compuesta usando #StableAudio

#InteligenciaArtificial #IA #tecnologia #arteconIA #stabledif...

Instagram

[serene jay]

nice! following! otro colega de España ya somos varios 😎 (yo en Barcelona)

[kind basalt]

Que dices! Ya me dirás quienes 🙂 ganas de hacer comunidad!

[shrewd crown]

Stable Diffusion with IMG2IMG and control net https://www.youtube.com/shorts/viVd668Z6JY

[fallen girder]

"Nocturnal Haze Hideaway" | Deforum-to-AnimateDiff vid2vid
https://youtu.be/CRx59PAV9qs

[supple estuary]

I use ControlNet reference mode with one pic and gif2gif. Why i have this effect? Its like 1 frame stuck

[shut pilot]

https://youtube.com/shorts/LxoLZOe_efM?feature=share

[noble bison]

https://youtu.be/i-ACONv3XmI?si=94nRAgqGYo42mhUj

[mossy ridge]

Hi! How could I generate only images and not animations?
I keep using the dream Prompt command but Stable diffusion generates animations instead of images!!

[shut pilot]

video, music, text and speach from an ai, for not so stable result https://youtube.com/shorts/ui2ZB9eIjy8?feature=share

[mossy ridge]

Hi! this week I'm not able to generate pictures but animations, in stable diffusion.

I keep using the dream prompt but it generates animations. do you know how to solve it????

[sage yacht]

Where are you generating?

[mossy ridge]

in one of the Bots (bot-1 etc...)

[sage yacht]

Right now generations are randomised between still image and video if not specified within the prompt

[trail marsh]

Hey everyone. Wondering if anyone knew where I can find a video or post about how to use vi2vid with automatic 1111 and animated diff

[shrewd crown]

https://youtube.com/shorts/eoGvlzkZJ8Y?feature=share

[snow creek]

I am not having any motion in animate giff first try,,..

[shrewd crown]

https://youtube.com/shorts/hBafvY6ckgs?feature=share

[stable hawk]

Made with AnimateDiff in ComfyUI. Using this workflow for single int input https://civitai.com/models/185421/hapasi

This was a sadtalker and tuneflow voice changer test

[native current]

/

[stable hawk]

AnimatedDiff in ComfyUI

[hallow zealot]

Here's a 1 min snippet of a video I'm working on, first run so yes I know it sucks...

[teal remnant]

Slightly mad...

[fallen girder]

qrcode_monster_v2 + hybrid motion, LCM 10 steps
https://youtu.be/jZsU8sHvTAg

[shrewd crown]

https://youtu.be/p66MD1eAmf8

[static niche]

anyone have a good SDXL LCM and the new SDXL Animdiff workflow yet?

[fast breach]

8 step video to video with animatediff+lcm - https://app.flowt.ai/c/u7Qg-2

[serene jay]

@peak moth @fast breach Please let's keep this channel only for sharing content or tips/questions related to animation and video. If you want to share links to your pages/communities please use the general channel, also bear in mind the https://discord.com/channels/1002292111942635562/1040299965647437905 regarding external links

[fast breach]

Was not aware, thank you

[serene jay]

no problem, we did not delete the comment since it was animation related.

[noble bison]

https://youtube.com/shorts/jfHOpxgLQ6A?si=3cdYTlRinpCcmx9X

[quasi spire]

When this happens 1. remove the controlnet, 2. render an image ( you can cancel, just get it started), 3. re-add controlnet, 4. render your stuff. It happens.

[novel meteor]

https://www.youtube.com/watch?v=FehfaLbTUjk&t=665s Hey!! so im using this video to make an AI beat reactive visualizer for a class project and I was wondering if ya'll have any advice on best ways to prompt / best practices to follow? Any and all help is appreciated!

[simple jasper]

This collab tool seems pretty dope! However, according to what i saw on Deforum with Music Sync, most peoples are using Parseq, a similar tool with different controls and UI. Here's the tutorials a community member gave me when i started to learn about it. I never tried the collab tool you just sent so i won't be able to say which tool is the more efficient, but feel free to check this out if you want to try using another tool https://www.youtube.com/playlist?list=PLXbx1PHKHwIHsYFfb5lq2wS8g1FKz6aP8

Hope that helps 🙂

[shrewd crown]

https://youtube.com/shorts/lC4X7SXYZLo?feature=share

[novel meteor]

Does anyone know how to use an init file with google colab on the stable difusion web ui?

trying to get a spiral beat reactive video made following this tutorial https://www.youtube.com/watch?v=mwWS8DyG_HU&t=127s

or is there a way to upload the video directly on mac?

[novel meteor]

Trying from the colab file source. will report if it works

[novel meteor]

it works

[normal oriole]

this is gorgeous

workflow?

[vital yacht]

AI Animation 3D | SDXL+Deforum (A1111 Webui extension)
https://youtu.be/nRs8ybdksXU?si=rhvEQ4GE4I1SAsyZ

[ionic quest]

How to animate multiple images to a video?

[void girder]

https://www.youtube.com/watch?v=pfDKiQTe08c

[faint rivet]

https://www.youtube.com/watch?v=ottn-NtHr2Q

[shrewd crown]

https://youtube.com/shorts/O-rEcrmg88c?feature=share

[quasi spire]

https://youtu.be/IKGG4e4dHo8

[serene jay]

Hey everyone!, @eager nacelle is organizing an AI Film Festival in France, you have a few days still to submit your movies, check it at community projects> #1174716131613876265 message

[eager nacelle]

Thanks Wyz!

[fallen girder]

"Wired Nightmare" qrcode_monster_v2, LCM 10 steps (Deforum)
https://youtu.be/cH4u3anf-08

[stable hawk]

ComfyUI and AnimatedDiff Evolved using a single Int

[foggy imp]

Just throwing this out those for those who are interested in learning about animation/Blender, etc! https://youtube.com/playlist?list=PLbJAuQIEql3YuacjyEedREPnwgljH7CST&si=_CijBmZPmVB8uIhy

[noble bison]

https://youtu.be/UJoEm9-U_x4?si=ip1Mxc7Hs60mSwA-

[proven tartan]

Anybody have any advice for tools to get rid of the flicker in an ai animations? I see the top videos, even though they have morphing inconsistencies, they morph into those smoothly

[stable hawk]

Film has been pretty solid for me as of late. Automatic1111 has some tools for it, Deforum does, and ComfyUI has some custom nodes for it.

[proven tartan]

Does deforum take an already made video as input? I was looking to do that but didn't see a way

[stable hawk]

If you go to the last tab of Deforum, at the bottom of the page there's a section for film and slowmo. You can drop the video in there just to interpolate or slow it down or both. Rife and Film both are an option

[proven tartan]

Of it supports rife...I wish didn't just waste time installing it on the cli

Shit your right, I would have never found that without asking or a guide, it's all the way at the bottom there

[stable hawk]

Yeah lol, it's kind of hidden and tucked away. Rife is finicky though, sometimes you'll get an error with rife, but if you try to Gen with rife, then cancel it, it will usually let you use it to interpolate after.

[proven tartan]

@stable hawk you wouldn’t happen to know what the comfyui nodes for interpolation are called?

So I don’t have to switch between the apps but just put it into the flow

[stable hawk]

ComfyUI-Frame-Interpolation and it's got a ton of options. I use it in most of my workflows.

[stable hawk]

https://github.com/Fannovel16/ComfyUI-Frame-Interpolation

[faint rivet]

https://www.youtube.com/watch?v=5G4XIgsg6no

[stoic fox]

Hi, here is a music video I made
https://youtu.be/-rsoqy6dXxk?si=WXLYU9l3yOlsQt2z

[void girder]

https://www.youtube.com/watch?v=O27_PCqN6eM

[old timber]

Hello, does anyone know about deforum "guided images"? The problem is only squares images work I guess.. ? I made a test with the same images/prompts with equal width and height so guided images works and I tested with a bigger width than eight, so guided images don't work and I have an error : "images not match". Why? Can I make guided images with landscapes view? Guided images is the best way I found now to use my own images. Thanks a lot

[quasi spire]

shorts version of I Drive https://youtube.com/shorts/K6LtnYXpLMA

[quasi spire]

https://youtube.com/shorts/sA-CkEW0Ews

[prisma osprey]

Hi there! We are looking for AI artists (animation) for a couple of projects, send your info and samples of your work at [e-mail]

[stable hawk]

vid2vid ComfyUI MakeReal lol... https://civitai.com/models/204423?modelVersionId=230161

Added wrong screenshot the first time lol.

[quasi spire]

reActor GUI + Blender , full video version https://youtu.be/_tUtMOB1y8A

[stable hawk]

using different loras with my vid2vid animatedDiff Workflow.

[proven tartan]

Any tricks in animated diff to get the perfect loop? Do generate more frames than they need and the handpick the most consistent ones?

[stable hawk]

It varies, if using an image int I play with the IP adapter strength, additional noise, and etc while adjusting prompts. I use Loras often to make things easier, then I'll use videos I've made with 3d animation tools. It really depends. It varies a lot

These are all just vid2vid ones so the initial video carries most of the weight.

Like these all came from the same video int with Different loras.

[proven tartan]

Thanks for the advice

Btw you wouldn't happen to know how to do the control net inpainting in comfyui?
Not sure how to apply the inpainting control net

[white gate]

awesome

[stable hawk]

I actually have no idea how inpainting works in comfyUI 😅. Automatic1111 has openoutpaint and I pretty much use that exclusively for my inpainting and AI editing

[serene jay]

https://github.com/taabata/LCM_Inpaint_Outpaint_Comfy

with LCM, it's really fast

[toxic glen]

@stable hawkthought your name was really cool, tried it out with a ghost lora

[thorn hare]

hey there, I am trying to use AnimateDiff with a Custom Furry model, but it always ends up totally garbaged, and I cannot figure out the sampling settings that does not break the rendering process. I am using a Laptop from 2014 and a GTX970M. Is AnimateDiff generally not possible with this hardware?

[serene jay]

uuh that's a tough one, animateDiff is vram hungry, maybe there's some hack you can do with LCMs and whatnot let's see if somebody else knows workflows for older hardware

[thorn hare]

Yeah I read that, having 6GB here :/ I also read that it requires at least 512x512 resolution but I cant go over 256x256 without it breaking apart

that is, 256x256 with 2x upscaling because the Checkpoint i'm using doesn't like 1x

Anyone having any suggestions for stable sampler settings on 256x256 maybe?

When I use the same settings than without AnimateDiff, it all comes out garbled

[serene jay]

I see, yeah with that low vram you need to do 256 and then upscale. What could be happening is that the SD 1.5 models are trained at 512x512 so they expect that, maybe there is a combination of sampler/scheduler that is not good at that resolution.

try something like euler/normal or euler/karras, that one is usually good for animation

upscaling only works above 1x, 1x doesn't do anything, to upscale you have to 1.5x, 2x etc

[river pelican]

Using SSD-1BA111 and DPM++ 3M SDE
https://youtu.be/LUMMRF8QPFQ?si=eh2SxqeijOATCHvO

[idle sentinel]

https://youtu.be/5WdavIUcf08?si=CbF44Soxt2NBrQ_C

This was super challenging, my first time using sdxl model, first time using controlnet, my first time training loras and my first time using deforum.
I learnt so much making it so I’m looking forward to doing more!

[silent galleon]

https://www.youtube.com/watch?v=cVUJmsVkUW4

https://www.youtube.com/watch?v=0XtoHqkFTmY

Using SD 1.5

[fallen girder]

"The Vortex" Deforum Hybrid Animation (qrcode_monster SDXL)
https://youtu.be/EwxGCHq9Ifo

[simple jasper]

So there's gonna be a Stable Video Diffusion tool now?! Really stocked to see what it uses, if it use some AnimateDiff, Deforum, Ip adapters, LCM or if it's fully dev by Stability 👀

[quasi spire]

those ip adapters... i need to get on comfy....

[quasi spire]

without them I'm stuck in Hollywood https://youtu.be/38dnNUBKT0g

[static niche]

What's the skinny folks? Seeing some tweets with sdxl animdiffs but no workflows yet. Anyone have luck getting it setup?

[serene jay]

https://comfyworkflows.com/animate-diff

you can toggle just the SDXL switch and there's quite a few

[static niche]

dont see any using the new SDVideo model unless i'm going blind, which i might be

[serene jay]

You said sdxl, not sdv 😜 anyhow on the #▶｜stable-video-diffusion channel there’s a comfy workflow people are working on

[static niche]

hah my bad 🙂

you're supposed to know what i was thinking and not what i said.. 🙂

didnt see that channel, never scroll up that far on channels list anymore, thanks!

[serene jay]

my bad I'll try to get better at mind reading 😜

[serene jay]

got you bro #▶｜stable-video-diffusion message

[static niche]

think it's gonna be awhile yet hah, these are coming out weird

[noble bison]

https://youtu.be/nbsWpqgF0jc?si=gpa_7l2N0O1ZG8iB

[void girder]

https://www.youtube.com/watch?v=Kri9Va_4530

[quasi spire]

I don't know where i should put faceswaps, so they'll go here... Having too much fun injectiong myself into clips!

[quasi spire]

https://youtube.com/shorts/L9xW1jvG-zk?feature=share

[toxic glen]

If you want to learn more about AI animation and AnimatedDiff techniques, come on by the CivitAI office hours where jboog teaches his top tier animation styles https://www.twitch.tv/Civitai

This is where I pick up the best tips and tricks anywhere on the internet

[silver island]

are the spec requirements for stable video diffusion different than the ones for images?

[static niche]

I've maxed and crashed a pair of 4090 20 gb gpus just trying to generate 4 seconds of video with no loras. 3 seconds is about top at least for the time being.

[silver island]

ah okay

so it sounds like its not even doable for consumer cards and regular people

[static niche]

Ugh stupid autocorrect keeps changing 24 to 20. But yeah it's a steep requirement to get anything decent yet

[silver island]

ah, good to know

[vapid skiff]

FAQ: How do I generate images? Is there a bot on the server?

Currently, there is a public bot on the server that generates images available as a research beta for SDXL, you can find the current status of the bot in #1047610792226340935. There are plenty of ways to use Stable Diffusion such as the official https://dreamstudio.ai/ website or by running Stable Diffusion locally using your own hardware - check out #1080946152318443610⁠ for more details! You can also stop by #1025467151206854736 for any issues you experience while using DreamStudio or #🤝｜tech-support for any problems you encounter while installing it locally!

[serene jay]

Between the guys at #▶｜stable-video-diffusion and comfy, requierements are down to around 10-12 GB, a lot more consumer cards are working fine

Even 8Gb are working (few frames)

[static niche]

They work fast! Awesome

[toxic glen]

First stable vid

[void girder]

https://www.youtube.com/watch?v=N_3p7FA5hFs

[shadow herald]

https://www.youtube.com/watch?v=NN8jfMZVzZ8

[hoary escarp]

Make music

[noble bison]

https://www.youtube.com/watch?v=drUsc1Vfy6s

[molten spire]

1200 BMW in 39 sec with deforum. Model: Dreamshaper and cartoon models merged (I cant remember settings sorry).
Prompts:

{
"0": "BMW m3 e21",
"300": "BMW m3 e30",
"600": "BMW m3 e36",
"900": "BMW m3 e46"
}

Positive: neon lights, masterpiece, high quailty, car

Negative: nsfw, nude, bad quality, worst quality

[serene jay]

Visuals experiment. Prompt travelling over AnimateDiff. Model: kakigoriV2 by @plucky shale https://youtu.be/ZgUomyIz9wE

[silent flame]

https://twitter.com/PUBLIKFRUIT/status/1729123610323808325

[fallen girder]

"Lost Signal" (AnimateDiff)
https://youtu.be/AECt5KV5f9E

[void girder]

https://www.youtube.com/watch?v=jKt1caBdf80

[vital yacht]

100% Deforum (3D mode) diffusion.
No interpolation/deflickering, No Controlnet, No Image/Video init

🍄 https://youtu.be/O5Vh9qnJoiM?si=zIYdmWPtf-9MD_Ou

[full flame]

#1098959457964277831

[forest escarp]

https://youtu.be/VaKlKcEZc4o?si=mDtREeanC9YZUb7I

[faint rivet]

I present to you “JFK Overdrive” the half-speed version of the octopus fight... Full speed coming in another video. But this one though... You get to see the details, and you get "The Star Spangled Banner," "Ride of the Valkyries" with JFK Inauguration Speech over the top. Somehow, I think it works. https://www.youtube.com/watch?v=EQV5BzAqwWU

[faint rivet]

So, here's the full speed version of my last post. But, this one has music that I wrote the lyrics for and suno.ai performed! That thing is so crazy.
https://www.youtube.com/watch?v=jV9opRehHnc

[shrewd crown]

https://youtube.com/shorts/JgNpdbEJnec?feature=share

[covert thicket]

https://humanaigc.github.io/animate-anyone/

[faint rivet]

https://www.youtube.com/watch?v=-v6oQPA02yU

[void girder]

https://www.youtube.com/watch?v=aRI2IR98Hh8

[faint rivet]

Good things are in store... https://www.youtube.com/watch?v=Kqgj23dSnBA

[dark sage]

https://youtu.be/i-OACGb1LLY?si=rcux6gtwj_kc50dp

[balmy crag]

@woeful tinsel

[mossy ridge]

new video for my song Walking Away

[serene jay]

https://youtu.be/Uhz7Dbpz5Qc?feature=shared

[faint rivet]

https://www.youtube.com/watch?v=DN3OV14MWZg

[stable hawk]

Testing out some stuff. Sad talker, SVD, AI voice changer from Tune flow.

[ionic zenith]

Holy hell... 7 months in the making, finally done with this video for my song Stranded.
I experimented with some techniques here, like layering the video input with its vid2vid output and messing around with opacity. It's a bit of a challenge to work with but when it works it's pretty cool I think. https://youtu.be/YvNEUUy2bKU

[simple jasper]

Wow crazy !

Anyone tried the new MagicAnimate thing ?

Or AnimateAnyone but it's not public yet

[mossy ridge]

hello, does animatediff work with sdxl?

i got to install the animatediff extension and donwloaded the mm_sdxl_v10_beta.ckpt into the animatediff model folder, but when i run the prompt i get the following message.. altho i see it generated 16 images but no animation.
what is this about ... i downloaded that from huggingface page

*** Error verifying pickled file from C:\StableDiffusion\stable-diffusion-webui\extensions\sd-webui-animatediff\model\mm_sdxl_v10_beta.ckpt
*** The file may be malicious, so the program is not going to read it.
*** You can skip this check with --disable-safe-unpickle commandline argument.

[void girder]

https://www.youtube.com/watch?v=gk9ulqfKVFU

[faint rivet]

https://www.youtube.com/watch?v=94d3-YNUXjY

[fallen girder]

Made with SDXL and Stable Video Diffusion on ComfyUI, FILM interpolation, and Topaz upscale. AI Generated Music with Suno.ai
https://youtu.be/7l8M5IE07DI

[meager otter]

Bright dandelion covered with frost at sunset in backlight voque style, hd, octane render, shadows

[sage yacht]

#1072220168534642768

[modest compass]

Made with SD 1.5 in A1111 with some controlnets and the LCM-Test Sampler: https://youtu.be/d1WmtB0f1yU

Made with SD 1.5 in A1111 with some controlnets and the LCM-Test Sampler: https://youtu.be/y0K61GHqq9s

[gaunt void]

AnimateDiff in ComfyUI with controlnets is pretty neat

anyone know of how to further improve temporal stability?

[modest compass]

Made with SD 1.5 in A1111 with some controlnets and the LCM-Test Sampler: https://youtu.be/IBW1_5tXB9Q

[fallen girder]

"Shrooms" (Deforum / AnimateDiff)
https://youtu.be/LJP8dq5hykM

[modest compass]

Short Clip from the german comedy movie "Papa Ante Portas" - in the Train. Done with StableDiffusion 1.5 and A1111 and some controlnets. https://youtu.be/8QrL43omWKY

[pastel rock]

https://www.youtube.com/watch?v=45iRGQM0GJo

[modest compass]

ChristmasVacation 01A

[faint rivet]

https://www.youtube.com/watch?v=I5y3WyzQ3Pw

[stoic fox]

https://www.youtube.com/watch?v=b4L9arGhfmI

[fallow surge]

All that effort and botched the title image.

"of", not "off"

[ionic zenith]

Playing with transparency layering a vid2vid with its source gives cool results sometimes.

Full vid: https://youtu.be/YvNEUUy2bKU?feature=shared

[mossy ridge]

https://www.youtube.com/watch?v=TlN2PNH_Nec

[analog fjord]

how in the world did you create this? this is amazing

[faint rivet]

https://www.youtube.com/watch?v=ICOX3stgUdU

[void girder]

https://www.youtube.com/watch?v=JDcz12_gYOA

[sinful lotus]

how are people getting more than 16 frames with animatediff?

[sinful lotus]

nvrm

[livid rampart]

how can iI improve the quality

less pixlation

here is another one i just want o improve the pixel, make more sharp

[sinful lotus]

more inference steps?

[livid rampart]

this one is through topaz ai, i there a way to restore facial features or only way is in post?

[fallen girder]

"Eldritch" Deforum prev img flow, SDXL bad trip, music made with Suno
https://youtu.be/2HuDo9icIlk

[spare goblet]

fantastic 😄

[faint rivet]

https://www.youtube.com/watch?v=Q89WrVnjqM8

[vital yacht]

100% Deforum (3D mode) diffusion.
No interpolation/deflickering, No Controlnet, No Image/Video init
https://youtu.be/KlSBRcQyZng?si=1zyXYejUgs9LCxFP

[fathom compass]

whats the best stable diffusion animation plugin rn

[fallen girder]

"Mourning" Deforum animation deflickered with AnimateDiff (TemporalDiff v1 + ControlGIF CN), music made with Suno
https://youtu.be/wfm5av9QWAI

[modest compass]

Christmas Vacation -- Shopping Mall -- https://youtu.be/0efe4ynHmTE

Christmas Vacation -- Snake -- https://youtu.be/LNRJegn7VYs

Christmas Vacation -- Silent -- https://youtu.be/yrUFSTy7WCo

A dance video --- of course --- 🙂 https://youtube.com/shorts/H7KHCjMtk9c

[modest compass]

Really Consistent... StableDiffusion with Mistoon Model and Controlnets: Mediapipe Face, lineart, temporalnet.

#RunDiffusion #stablediffusion #stablediffusionanimation #stablediffusionart #consistentanimation #aiart #generativeart #midjourney #ai #machinelearning #experimental #deforumstablediffusion #animation #moving #visual #fyp #xyzabc

[shrewd crown]

https://youtube.com/shorts/nHGAWkqNWGs?feature=share

[faint rivet]

https://www.youtube.com/watch?v=UujcJx0hqIw

[strange moth]

Jellyfish try to escape !! 🪼

[digital ember]

Has anyone built Stable Diffusion "Animation" in to their app? Like hosting themselves with their own private API in the cloud, and letting users upload images and animate them? And maybe string a few together to make like a short cartoon?

[void girder]

https://www.youtube.com/watch?v=6jPgXFeSlaM

[inner crest]

Is there a way to remove the watermark from runaway ?

And how to improve the quality of Stable Video Diffusion ?

[white vault]

they don't like you to do it but a simple crop and distort will do

[faint rivet]

https://youtu.be/RHhmWXkF_yU

[covert perch]

https://youtube.com/@aliefrat?si=F3BbQnTc5YaEnvig

[fallen girder]

"Hypnosis" Deforum hybrid motion + qrcode_monster, deflickered with AD + ControlGIF
https://youtu.be/ojVgJ6CFhoE

[wispy stag]

https://www.tiktok.com/@yooartifical/video/7313931902656728363?_r=1&_t=8iHu0oBJ6Pj

[white vault]

https://youtu.be/y-_PuGSc4O8

[fresh orchid]

Music video with composited visuals, which include a Stable Difussion/Deforum animation clips. The song written, sung, and produced by me. All graphics are made by me also. I hope you enjoy it.
https://youtu.be/wJmQGVPLong?si=ger4FPEXmvR_79dD

[high rose]

https://www.tiktok.com/@dairydisaster/video/7314131765252263214

[fallow thorn]

Hey, slight bit of noob question here, but anyone have advice on getting better faces out of video like this where the human subject is in the middle distance? I'm using SD 1.5 in Comfy UI, openpose, canny, and depth controlnets, IPadapter, and Free U. Any loras that can help with this?

[wispy stag]

https://www.tiktok.com/@yooartifical/video/7314228549194353962

[runic breach]

a video of christmas festival

[sage yacht]

#1072220168534642768

[wispy stag]

https://www.tiktok.com/t/ZT8HAj4GV/

[halcyon mason]

https://www.youtube.com/shorts/pdoIabqt0xw

[wispy stag]

https://www.tiktok.com/t/ZT8H9Vrbm/

[wispy stag]

https://www.tiktok.com/t/ZT8HQM8Tk/

[fallen girder]

Deforum deflickered with AnimateDiff + ControlGIF, Music by Suno
https://youtu.be/xWb902CcDg8

[sand sonnet]

Is the gradio ui for the animation SDK broken? Went to my old project after about a month and I get "engine ID not found" for every model I try.

[nocturne wharf]

made an autonomous short film, using GPT, musicgen and animateDiff v3: https://twitter.com/fabianstelzer/status/1737585786843787495

[faint rivet]

https://www.youtube.com/watch?v=a3-IKJBSgUM

[noble bison]

https://youtu.be/YAVG_FkAiR4?si=lm19AaD4F4fOF3cX

[wispy stag]

https://www.tiktok.com/t/ZT89NWtN2/

[granite talon]

||heleo||

[ionic zenith]

Cardboard over AI skies.
AI faces over faces.
AI over AI.

https://youtu.be/YvNEUUy2bKU

Two of my favorite screenshots.

[wispy stag]

https://www.tiktok.com/t/ZT89MNsWN/

[noble bison]

https://youtu.be/7e7zkEnrkdY?si=XWN04PnbeuS6VNal

[mossy ridge]

https://www.youtube.com/watch?v=NBRZhL8qFT4 als comic und mit neonfarben

[wispy stag]

https://www.tiktok.com/t/ZT89u4QKP/

[void girder]

https://www.youtube.com/watch?v=feFS70kQG20

[sudden cargo]

Hi sorry if this is not for questions and just results but can anyone direct me to a tutorial or tell me what technologies I need to install in comfy UI to render animations?

[wispy stag]

You asked me earlier but I wasn't sure how to reply because there are many tutorials you have to go through and I dont have a source for them. I use animatediff, controlnets, loras, and other tricks like lcm for speed.

[sudden cargo]

Oh okay yeah I asked but I figured you didn’t see it haha dw… this sounds good tho I will research how to get controlnet in comfy first since I haven’t gotten around to working with that, only with impact for adetailer.

Thanks for the info tho !! Also love your animations haha

[sudden cargo]

@wispy stag do you use small 1.9 go models to generate each image? Also how many images should I generate for a single 30 sec video at idk 12 frames per second

[wispy stag]

idk what 1.9 go models are lol

and I do 300 batches and it takes about 6 of them for a 30sec video (60fps)

[sudden cargo]

Oh my bad I meant 1.9 gb models are you use since those seem to be faster at generation. Or does it work equally well with larger models ?

[wispy stag]

https://www.tiktok.com/t/ZT89VEG7B/

I made this today took 2 hours to render

[sudden cargo]

Damn; if it took two hours for you I would imagine for 8gb vram and card will take at least double or more

[fading dagger]

Guys, an off topic question, now I've seen around the internet how videos for AI influencers work and they all seem very poorly made. Do you know any methods to do them better? Thank you !

[white vault]

https://youtu.be/HqyWlZWWrGs

[wispy stag]

https://www.tiktok.com/@yooartifical/video/7316128002037894443

[copper shell]

sup, im trying to use animateDiff on a1111 and i can't make it work. It just generates almost still images as videos. And the quality is way worse than the very same prompt and seed used on a regular image. Any clues? i'm using controlnet tile, and now i'm trying with prompt travel since i guessed maybe it doesn't move because there's nothing making it move, like, say.. a change in prompt. Any help would be super welcome. And before you say it, no, i won't use comfi UI at least for the time being

[serene jay]

https://youtu.be/YaBZzAiZuos

[copper shell]

it worked now, somehow. Not sure what of all the things i've tried worked, but at least now i know it works

the quality is still a bit crappy, like LCM with 1-2 steps crappy. If anyone has an idea on why it happened despite using 20 steps...

[sudden cargo]

Can anyone direct me to a good guide or custom nodes github page from where to get controlnet inside of comfyUI?

also should I download the 1.1 models for all controlnets or use the already downloaded models from the webui installation that say 1.5?

[wispy stag]

Lmao same guy who told me I didn’t know anything asking all the questions I have answers to though 🤣 stop choosing a processor and use a reference image.

[wispy stag]

Use “comfyui manager” to install custom nodes without the need of github.

[sudden cargo]

Okay I got the new fork of contro net preprocessors, jt says the old one is no longer maintained! Rn im going to see how long it take on my gpu

@wispy stag , quick question,do you know why it says its downloading models on its own like, every node downloaded its own model when I connected my stable diffusion models for the controlnet path for comfy

is this normal?

[wispy stag]

That sounds nice tbh lol I had to manually add everything

[sudden cargo]

oh but the OpenPOse Pose Recognition is working ummmmm weird, its like there are two nodes that were added

1. OpenPose Pose Recognition which doesnt allow me to check which Model I want

2. DWPose Estimation which gives me a BUNCH of onnx models that work even better than the other,

am confused since in the tutorials I saw, the normal open pose was supposed to work perfeclty

[wispy stag]

there are a few Open pose models

gotta use the right one

[sudden cargo]

may I send the image here?

[wispy stag]

I forget the name, I renamed mine to OP 15

[sudden cargo]

I literally have no idea what I am doing rn, webui did it for me all the time haha

[copper shell]

i don't want any answers from you with that attitude. Also i already solved the issue by myself. Get lost

[wispy stag]

You’re the one with the poor attitude towards me first bud 🤣

[wispy stag]

https://www.tiktok.com/t/ZT8xdPHCv/

[wispy stag]

https://www.tiktok.com/t/ZT8x8Lu2X/

[white vault]

🧠 👇🏻 'brAInz'
https://youtu.be/r-NST4ta_kQ

[faint rivet]

https://www.youtube.com/watch?v=nw8wgeVik2g

[wispy stag]

https://www.tiktok.com/t/ZT8xhur5d/

[fallen girder]

Deforum + AnimateDiff / controlGIF
https://youtu.be/fBK6pQQ2JTs

[strange moth]

Kendrick ? ☁️

[shrewd crown]

https://youtube.com/shorts/Y-w87l7NyX0?si=DLjRXBDAbMDdVU2T

[faint rivet]

https://www.youtube.com/watch?v=2sgWCz-xbJk

[faint rivet]

https://www.youtube.com/watch?v=njVNbLH2Jo4

[wispy stag]

https://www.tiktok.com/@yooartificial/video/7317167251285740846?is_from_webapp=1&sender_device=pc&web_id=7285747949094585902

[tranquil ingot]

https://www.youtube.com/watch?v=caH6aXkG5DY

https://www.youtube.com/watch?v=fsrguHWUYAI

[wispy stag]

https://www.tiktok.com/t/ZT8xpmhsc/

[mossy swift]

Hello, I am using automatic1111 (if it makes a difference) and was wondering what are some good videos/tutorials on making animations. I specifically want to make animation using recordings/existing videos but, am open to anything good. Thank you.

[fallen girder]

Deforum deflickered with AnimateDiff, music by Suno
https://youtu.be/nXk8wULhJnw

[wispy stag]

https://www.tiktok.com/t/ZT8Q2bX8t/

AI’s home and she’s ready to merge

[white vault]

https://vxtwitter.com/contrastruction/status/1740812790879994040

[strange moth]

nuclear cloud explosion / mushroom concept assembly with multiples mask inpainting, ControlNet, Multi IPadapter, animatediff

[wispy stag]

https://www.tiktok.com/t/ZT8QWn29Y/

[wispy stag]

https://www.tiktok.com/t/ZT8QTbMu2/

[inner crest]

AnimateDiff + Controlnet...

hummm... I start to understand how to use Comfyui and amazed by the speed and consistency of that thing

[noble bison]

https://youtu.be/XnK2f7zSkLg?si=p-KIBrL16A_pVSAa

[inner crest]

raaaah Im close !

[inner crest]

https://tenor.com/view/aladdin-genie-thats-how-genie-magic-works-magic-will-smith-gif-25252385

[safe oriole]

i reckon multiple ring ip adaptor masks could do it, not sure how with current ai you could do the bubble coming out without a broken down curve, maybe 5 ring parts

[white vault]

https://youtu.be/am2H_6Hchbc

[wispy stag]

https://www.tiktok.com/@yooartificial/video/7318687169294585134?is_from_webapp=1&sender_device=pc&web_id=7285747949094585902

[opal dawn]

https://youtu.be/GHql4qFD35Q

[strange moth]

Experimental 😵‍💫

[inner crest]

https://tenor.com/view/bh187-stargate-portal-gif-19287628

I think without a CGI tool, you cannot do the real vortex from Stargate

even if in the movie, it looks like they recorded a flushing toilet, and added it inside of the ring

So, it may be possible to get some kind of water vortex with ai, animate it and use video editor tool.

[vital yacht]

I created this video by combining 2 animations created separately with Deforum (WebUI A1111 extension).
100% Deforum diffusion, no interpolation/deflickering, no Controlnet, no Video Init.
HAPPY NEW YEAR!
https://youtu.be/THciD-RsLPU?si=FmKN0nBIvwMnbKeV

[wispy stag]

took me 10mins lol

could look better if I spent more time on it

[inner crest]

love it !

[drifting cedar]

when i try to start animatediff i get this error?

RuntimeError: The size of tensor a (32) must match the size of tensor b (24) at non-singleton dimension 1

someone know why i get thtis?

[wispy stag]

probably model, can't mix XL with 15 and so on

[dreamy frigate]

Happy new year guys 🎊
Here's an acoustic riff by AI generated me in different places around the world, to celebrate and inspire you 😁🤖🎸

I've created a new Instagram profile @ai_guitarist where I'll be posting more cool guitar videos, transformed by AI - follow if you are interested where this goes!

I'm still working on the model but at some point I could create an easy to use app to unleash guitarists creativity, if there's enough interest 🤔

https://www.instagram.com/p/C1kJQRxMAS6/

https://youtu.be/fzaKPdOUa0g

shefa

🌏🎸 New year, new AI-generated me!

A robot samurai under Japan's cherry blossoms 🌸, a robed wanderer in the Sahara 🐫, a cosmic performer under Iceland's awe-inspiring Northern Lights 🌌 and a tuxedoed musician in New York 🏙️  All from my room! 🎵

Is this the era of AI guitarists? 🤖🎶

Drop a comment if this chimera of music and technology amaz...

Instagram

[wispy stag]

https://www.tiktok.com/@yooartificial/video/7319483199212653870

[fallen girder]

"ElectroJunkies" Suno>Framesync>Deforum SDXL>AnimateDiff>Topaz
https://youtu.be/kN8jdvRQvzA

[inner crest]

They say it lead to another world...

[glad vessel]

https://www.youtube.com/watch?v=VOP6BcxsMds

[white vault]

https://youtu.be/t9gr69OkRPE

[stiff iris]

https://tenor.com/bXbSx.gif

[void girder]

https://www.youtube.com/watch?v=JfxevioMFT4

[fallen girder]

Suno > Framesync > Deforum > AnimateDiff
Full animation upscaled here: https://youtu.be/SO_G8eVirQ4

[hazy olive]

this looks so cool

[fallen girder]

Thanks!

[noble bison]

https://youtu.be/zwwY2edQwSQ?si=rkgEf4Ivk0K6Mk_V

[shrewd swift]

https://youtube.com/shorts/g9rfbQg14k8?si=6FvF64SbK4Y2kweY

[summer night]

тату мультик

[wispy stag]

https://www.tiktok.com/t/ZT8Xo6F8s/

[hazy olive]

"f this i´m leaving i don´t wanna be in a world with these TA people"

[near pagoda]

Steamboat Willie converted into an Anime. This was a poor attempt and my PC shut off (probably a Windows update) before I could finish.

https://www.youtube.com/watch?v=MohQSRC8Njk

[white vault]

further down the rabbit hole....
https://youtu.be/hDFO_OrCjLo

[wispy stag]

https://www.tiktok.com/@yooartificial/video/7321302445819170091

[wispy stag]

https://www.tiktok.com/t/ZT84pywFY/

[inner crest]

so beautiful...

and so consistent....

[inner crest]

sacrebleue !

[wispy stag]

https://www.tiktok.com/t/ZT8VLKDre/

[inner crest]

It would be cool to have in Comfyui some kind of switch. If I want to upscale, I click on it. because now, I have to mask the nodes I don't want

According to people on reddit, Stable Diffusion is just about pushing buttons ! So why I have to do all the work !

https://tenor.com/view/very-unacceptable-trex-gif-22252091

[wispy stag]

I can DM you a work flow with switches on those nodes, you toggle them on and off for ez

[inner crest]

https://tenor.com/view/pennywise-gif-27016910

[shrewd crown]

https://youtube.com/shorts/fwVVKtzKFRQ?feature=share

[inner crest]

I created a workflow

first video is raw from Animate Diff
Last is upscaled
would you be interested to have it ?
Or is it something easy to make ?

[wraith furnace]

(A moment of pastel-colored powder explosion is captured in stunning detail, frozen in time with a super long exposure and slow shutter photography technique) (In the foreground, the vibrant and delicate powder bursts in every direction, creating an ethereal effect) ((Behind this dynamic display, a wall features an image that echoes the explosion, itself bursting into life with multi-colored pastel powders resembling a jellyfish hovering in water)) (((The composition showcases a harmonious blend of delicate pastel colors, creating a sense of depth and texture))) (Lighting is meticulously balanced to accentuate the pastel hues and the fine details of the powder's movement) (The overall atmosphere of the image conveys a serene yet dynamic beauty, capturing the juxtaposition of movement and stillness)

Firecrackers are being set off

[faint rivet]

https://www.youtube.com/watch?v=Si4S7I8zcMI

[wispy stag]

https://www.tiktok.com/t/ZT8VvmXcC/

[fallen girder]

"Hallucination" Deforum + AnimateDiff
full animation upscaled here: https://youtu.be/WEHRMYwRlr0

[dreamy frigate]

Fly Me To The Moon - AI Guitarist Cover

Let me know what you think, and if you find a funny diffusion artifacts 🙂
Youtube - https://www.youtube.com/watch?v=odheRPBMDCg&ab_channel=AIGuitarist

[inner crest]

I release my workflow
https://openart.ai/workflows/elephant_present_36/vid2vid_animatediff-hires-fix-face-detailer-hand-detailer-upscaler-mask-editor/LwT4AB8wEtKQilrjq2G8

send me your feedbacks. Im new to this and not sure I made a good job

[inner crest]

And it is made from a workflow from @waxen basin . I only improve it to be better, faster, stronger !
https://www.youtube.com/watch?v=P4IdHKHrb48&t=1067s

https://tenor.com/view/another-one-dj-khaled-khaled-another-you-da-best-gif-5951590617028384526

[waxen basin]

yo! gonna have to try this! Thanks for sharing it 🙏🏽🙏🏽

[wispy stag]

Did you even watch it yourself? 🤣 the entire video is gray after the first few moments.

[dreamy frigate]

Oh, what? are you on web or mobile?

[wispy stag]

I tried on my PC first and I was like is it me? So I checked my phone too and its the same. So i'ts the video

[dreamy frigate]

wow it is like this for me

I think youtube f*cked it

compression or something

[wispy stag]

That's annoying and unfortunate

Sorry that happened

[dreamy frigate]

Hm, my friend says its fine for her 😄 what?

[wispy stag]

weird

[wispy stag]

I will try it out tomorrow and let you know

[dreamy frigate]

It's like this for me, I think I need to reupload

THanks for letting me know!

[dreamy frigate]

https://www.youtube.com/watch?v=anoGAah7RWQ&ab_channel=AIGuitarist

That better?

[wispy stag]

Yup working now

[dreamy frigate]

It's good that my channel just started so I don't lose anything for reuploads lol

[wispy stag]

https://www.tiktok.com/@yooartificial/video/7322943561861778734

I just finished this one

[vital yacht]

METALLICUM 🛠

Static images created with Stable Diffusion XL
Animation created with Runway Gen-2 img2vid tool

https://youtu.be/yx8wuCSx0tM?si=9sfIbZl-8yfGPIoH

[regal sage]

how? 👌

how you got this consistence result??

[wispy stag]

https://www.tiktok.com/t/ZT8qsNqNV/

[inner crest]

https://openart.ai/workflows/-/-/LwT4AB8wEtKQilrjq2G8#comment-qONm95XUz0Jcjbl6vZlu

damn people downloaded 1000 times in less than 24hours...

I should have sold it to Topaz. I kill their soft with that workflow...

But if it is not free, it is not Ai !

everything for the community !

https://tenor.com/view/chine-partia-gif-464767192641316396

[strange moth]

Audio Reactive composition 🔊 | Animatediff x TouchDesigner
(wait for the big bass 🥴 )

[safe oriole]

yea i saw it up on there and i was like hey i recognise that name 😄 , been browsing there trying to find an animate diff 3 spare control workflow but only seen this one so far which looks pretty cumbersome https://openart.ai/workflows/lemur_runny_14/sparse-controlnet-animate-diff-video-to-video-extras/1LleqvhTeMWYJHedJyoj

[wispy stag]

This only allows batches of 32

thats not very long of an animation

[safe oriole]

true, not my workflow, trying to find one 😄

[wispy stag]

https://openart.ai/workflows/grouse_athletic_95/animatediff-v3-workflow-with-ipadapter-and-controlnets/H5F2mm5LdYuMvq3JjOmU

this one is good imo

can do 700 batches

https://openart.ai/workflows/jerrydavos/v21-animatediff-controlnet-animation-refiner---lcm-part-3/blaad6IdilUH7yjXuR9H

This one is LCM so you can do Huge batches

[safe oriole]

yea the 700 ones are pretty epic, im looking for spare control workflow workflow

i found the 2 nodes they used in the one example i found, inside advanced controlnet just not sure all the ways people use it, to me it looks like itd work great doing 1920s cartoon animations but would need like a lot of inbetween images etc to build the story

[sinful sand]

I've generated a fresh vid2vid. Please share any suggestions or feedback..

https://www.instagram.com/reel/C1_r3UopTo7/?igsh=MzRlODBiNWFlZA==

manish_minglani

Dance moves in the realm of AI cyberpunk really opening up it’s potential!

Dance Moves @sophiatoppo
Choreographer @devasmitaa
Song: Tre Songz - Na Na
.
.
.
.
.
#ai #artificialintelligence #creative #techonology #art #dance #fusion #dance #choreographer #aiart #aicommunity realimation #discodiffusionart #midjourney #stablediffution #ainimation

Likes

109

Instagram

[wispy stag]

https://www.tiktok.com/t/ZT8b1wSLX/

[void girder]

https://www.youtube.com/watch?v=UIrjdLGZHpo

[inner crest]

damn it is intense 😄

a lot of steps are not necessary but I guess it cannot damage the output

[inner crest]

I keep testing my workflow.

Last time I tried without my workflow it was looking like this 😄 😄 😄

from prehistoric era to Star Treck...

https://tenor.com/view/himym-barney-barney-stinson-stinson-homo-erectus-indeed-gif-17040871

[fervent fractal]

https://www.instagram.com/reel/C2EUxGsJel-/?igsh=MXZzaHA5bWN0dzJtdg==

govind.ai.art

Do you want to see the real magic? @jonboogz

#magic #ai #aivideo #stablediffusion #art

Instagram

[inner crest]

https://youtu.be/MbS1YcKhyyA?si=RHhG39rLcbLJFfZM

[weak lark]

To get a stable pace of an animation, is it simply just set max frames to be multiplier of target fps?

[wispy stag]

https://www.tiktok.com/t/ZT8bf4sfy/

[wispy stag]

https://www.tiktok.com/t/ZT8bakWhN/

[frail lake]

create barbie image

[wispy stag]

https://www.tiktok.com/t/ZT8bXBwBo/

[royal zenith]

https://youtu.be/QXfmK5vRBL0

[wispy stag]

https://www.tiktok.com/t/ZT8b4B5B3/

[devout tulip]

https://youtu.be/sYzXUuVQwFQ

[inner crest]

Improved my workflow to change backgroud easily with ipadapter

https://openart.ai/workflows/elephant_present_36/vid2vid_animatediff-hires-fix-face-detailer-hand-detailer-upscaler-mask-editor/LwT4AB8wEtKQilrjq2G8

[safe oriole]

cool you went to the foreground background split 🙂

[wispy stag]

https://www.tiktok.com/t/ZT8gJodxS/

[mental cobalt]

Guys, I managed to use all the methods, except video to video, I'm using the custom nodes from Animated DIFF, but I can't generate anything, it gives me an error

[wispy stag]

one of your nodes says this and it's the one thats lit up red meaning the problem is that

delete it and replace with vhs video combine

[mental cobalt]

ohhh bro how did I not read it, it was in front of me the whole time, I'm going to test it

[mental cobalt]

I switched to video combine, but the error continues in the Ksampler operator, I'm not understanding everything is ok, Ksampler seems frozen, I can't even drag

[wispy stag]

change the node to a different ksampler

like advanced maybe

[mental cobalt]

I tried to change but the error remains the same

[mental cobalt]

Disappointing, I've seen hundreds of videos, it seems so easy, I've already installed several custom nodes, everything is up to date, I tested it on another PC, the same error remains

[serene jay]

I'd suggest to start with a basic setup and then build up the workflow best suited to your needs, step by step. Doing it that way it's best to learn and also to troubleshoot problems with specific custom nodes.

[mental cobalt]

I started from the basics, I managed to use all the "modes" img to text, text to img, img to video, text to video, everything worked, I'm just having problems with video to video, and it's exactly what I need

[wispy stag]

https://www.tiktok.com/t/ZT8g4e6R4/

[fallen girder]

Deforum + AnimateDiff
Full animation upscaled here: https://youtu.be/XvCGPEIbuEE

[vital yacht]

100% Deforum (WebUI A1111 Extension).
No interpolation/deflickering, No Controlnet, No Video Init.

https://youtu.be/YxpQE1DbSb0?si=Z_bgYJYxz2SGd1hN

🔹🔹🔹🔹🔹🔹

Audio: "In Eternity We'll Meet"
by Aakash Gandhi (YT Audio Library)

[normal apex]

Hey everyone
Check out my first try with stable video diffusion
https://www.instagram.com/reel/C2WfRHFtG0t/?igsh=MTdkMWxpN2tlMTczeg==

ehsand

I used stable diffusion to make videos based of the images. Then tried to edit the clips in davinci reaolve.
Well, I like my first try😊 hope you like it too.
#video #clip #musicvideo #ai #aiart #stablediffision #armagedon #endofworld #art

Instagram

[vital yacht]

"Steelcore Sunrise"
100% Deforum (WebUI A1111 Extension) + SDXL.
No interpolation/deflickering, No Controlnet, No Video Init.

https://youtu.be/7UqXqfZr6yc

[fallen girder]

"Wasteland" Deforum hybrid motion + qrcode_monster, deflickered with AnimateDiff + ControlGIF
https://youtu.be/MTz3bbqGpOM

[strange moth]

MWHAHAHAH finally lips syncing

[strong spoke]

Do you want to see the real magic?

[strange moth]

Lips syncing test 2

[onyx thistle]

Four Seasons

[inner crest]

comment tu fais le lip syncing ? mediapipe face ?

[strange moth]

en partie ça depends des cas, surtout softedge/lineart en vrai

[glad vessel]

Does anyone know why animatediff's motion module changes the look of my image so much?
It makes it so much darker and less like my subject. improved3dmotion_improved3Dv1. Because mm_sd_v15 sucks terribly and won't even keep a corherant image at all.

[serene jay]

are you on auto1111 or comfyUI? at comfyUI you have the "color match" custom node, you can pipe your original to the animadeDiff result and it will color match them

[livid wasp]

Space Dunk - based on @djstephens30 dunk 🔥

https://youtube.com/shorts/LnbnDBUyvIY

[inner crest]

Submit your work !
https://aiff.runwayml.com/

[fallen girder]

"Sativa Dreams" Deforum hybrid motion + qrcode_monster + Framesync
https://youtu.be/qqzs_FE0V-I

[weak mirage]

very amazing lioncock

[shell flare]

thank you

[weak mirage]

reminds me of harold from fallout

[shell flare]

I've got 100+ random animations that I post on the Civitai discord and site, but I'm looking to branch out a bit. I may get banned over there for picking fights with Loli posters.

[weak mirage]

give up on that. they run the site. civit will burn because they facilitate those dudes so hard. its only a matter of time.

make sure you put your logo on those images if you're going to publish. too often people will just reshare like they made this stuff

instagram and x is almost a better sharing platform than civit could ever be. you know what the civit audience is

[shell flare]

Yeah you're right I gotta start watermarking

And Civitai, well, we'll see. The alternatives aren't much better

Insta sounds like a good idea though

[rugged willow]

i never saw you digging at people but ide probably support it lol, sorry you got banned hah

oh oh "may get banned", carry on

[weak mirage]

I wouldn't go to 4chan and tell them to stop posting pervy anime. That's how i feel about civit most times

[shell flare]

Hey dude

Yeah I may get banned but we'll see.

[glad vessel]

It's not only color, it's not the same image, look at background too.

[shell flare]

Looking fwd to the day it happens because I'll go down in a blaze of glory

[serene jay]

try tweaking your denoising settings

[serene jay]

what do you mean by this? this server is strictly NSFW, maybe some of the anime content could fall into "lolis" but it's mostly safe and when it's not, they go to horny jail for a bit ; )

[noble bison]

I used vid2vid on a video created with Mandelbulb, producing two versions: one with controlnet and the other without, keeping all other settings identical. I then edited the second video to include "GhostGlitch" effects, particularly liking the girl/ghost imagery https://youtu.be/KCJiZ3N9bis?si=JcexlOKefzHqV9Af

[wispy stag]

Um if tiktok allows this video please do not delete it here 🤣

[serene jay]

sorry mate, NSFW

[wispy stag]

Tiktok doesn't even think that lmao and their restrictions are steep

Whatever 🤣

I'll just put them in Utah Mormon dresses for now on 🤣

[frozen storm]

we aren't tiktok

[serene jay]

no need for extremes, 99% of the content posted here is OK, that one falls into sexualized poses https://discord.com/channels/1002292111942635562/1040299965647437905

[glad vessel]

In the controlnets?

[serene jay]

are you doing vid2vid? at the KSampler level running lower noise levels make the image look more like the ref. a 1.0 value is 100% generated, 0% value means that no diffusion is happening, the image in = the image out, you make balance between those 2 extremes

0.3 - 0.7 levels

but I don't know your workflow, you might be doing something completely different

[serene jay]

oh gotcha! do you need that advanced ksampler? you don't seem to be using the start_end features

use the reg one

and this 1.0 denoise value, use it at 0.5 and test

then finetune the level, maybe for your case it's 0.3 or 0.8

[serene jay]

another thing I'm seeing here is that you are using LCM but with the "normal" scheduler, try with the "sgm_uniform" which seems to work better for LCM, at least to me

[shell flare]

Stick around for the whole thing

It's supposed to loop

but you know, discord and videos.

[serene jay]

that's cool

[shell flare]

Thank you, this one def came out sick af

most of the shit I post here will be like 16-30 seconds or so

[serene jay]

post ! what's your workflow? is this sdxl based?

deforum? animateDiff?

[shell flare]

All animatediff comfy, super basic workflow

hald

there you go

@serene jay

[viral kiln]

👍

[shell flare]

fuck it, right?

[manic cargo]

somehow I cannot login Runway

for a couple of days

[serene jay]

Seems like something for runway’s server or support! 😁 having said that just logged into my account with no problems! Contact support

[manic cargo]

Thanks. Probably because I am from China. But I am using proxy...

[stiff cairn]

I want to use img2img to generate an animated image from the original image. Is there any plug-in that can do this?

[serene jay]

There are multiple ways to do that, where are you working? automatic1111?

[inner crest]

Wanna some vid2vid resource ?

[sterile valley]

the pharaoh in front of the pyramids, at his side a being with the head of a man with horns curved backwards and a hairy body and winged elephants guided by Satan the cursed in front of a people of frightened slaves... HD

[limber oasis]

In a charming village, Little Kai and Little Ellie play joyfully, hand in hand.

[crude solar]

AnimateDiff + Wav2Lip Studio

[stiff cairn]

Yes, I'm using automatic1111

[halcyon olive]

yo how long did it take to make this

[hybrid lily]

https://youtu.be/1uISXcgIW-c?si=V2HIVzlfFg17yYzb using parseq

[crude solar]

Calculation take about ten hours

[vestal nexus]

https://www.youtube.com/watch?v=VGUbdBvgGlQ

[rapid willow]

Can someone help me out with this SD animation API problem — When using xl-1024-v0-9 for generating keyframes, all works well (despite being utterly slow), but when I switch over to xl-1024-v1-0, the first frames until diffusion_cadence_curve = "0:(5)" → in this case the 5th, are rendered correctly but then quickly start fading to gray, what looks like 100% noise.

[serene jay]

have you tried this at #1098959457964277831 ? that channel is specifically for the api platform

[rotund snow]

What's your preferred upscaler? Are you using the native SD tools or a separate app? I've been rendering a video at 768x432 and would like to upscale it to a proper HD size like 1280x720 or 1920x1080 for YouTube...

[wispy stag]

https://www.youtube.com/watch?v=Y-RCYPS3Kuw

Wow this is the best AI music video I've seen

the resolution of 21:9 alone is crazy lol

[fallen girder]

https://youtu.be/0W4LRvQvaS4

[hoary kiln]

study

prompt variations

[frozen storm]

@wispy stag we see them. Thanks for your positive fueled msgs

[wispy stag]

[magic plaza]

✅

[rich gale]

Thank you #▶｜stable-video-diffusion diffusers from the Peter Gabriel challenge! You gave me a great start here🥰

[wispy stag]

https://www.tiktok.com/t/ZT8nn5aob/

[white vault]

https://youtu.be/oHBPWVnqSMI

[woven socket]

Eltena is young girls selling precious crystal dresses white tshirts big shop big place crystal potery with gold component

[wispy stag]

Sir this is wendys

[livid wasp]

Giving some fire 🔥 to one of the coldest knockout in UFC ❄️ by Joaquin Buckley

https://youtube.com/shorts/NuS8PPnv3D4

[wispy stag]

lmao nice the 2nd transition killed me 💀

[wispy stag]

https://www.tiktok.com/t/ZT8WFWTEm/

[rotund snow]

https://youtu.be/MTJKHvfhbqI

[royal zenith]

https://youtu.be/1_oM0Gqh1D4

[pulsar token]

This is absolutle crazy! 😄

[woven socket]

Eltena selling precious crystal with goold elementen in big salon with white pavment and walk arownd objects off crystal and muve them very precious and delicate object girl will look lik photo i upload

[reef igloo]

Hey, everybody.
https://www.youtube.com/shorts/Q3trHohFVDg

[onyx thistle]

Hello, anyone here experienced with animateDiff? I am creating a video with 3 controlnets for depth, lineart and temporalnet (they are all fed with folders that contain a series of pictures in the respective format) and the background changes alot altough the camera is not moving and it should be rather static, you know how i can make it change less?

the background in the frames im feeding isnt changing at all

[onyx thistle]

oh i think it was the depth map, i turned it down now its much better

[hoary kiln]

https://www.youtube.com/shorts/jn5n-64y964

[glad vessel]

https://www.youtube.com/shorts/A1a6Fl7T_p0

[onyx thistle]

may i ask about your workflow?

this looks really incredible

[hoary kiln]

sure, it's in my twitter

[onyx thistle]

Olivios vid? I think I tried this but it didnt work as well :S

I will try again, thank you!

your videos are really gorgeous...in what resolution did you render? and what GPU do you have if I may ask

[onyx thistle]

and what checkpoint were you using for this?

[hoary kiln]

custom merge of ChilloutNi + Majicmix. almost 50/50

but any model will do with right weights

[onyx thistle]

Thank you! You're a blessing

my stuff really doesnt come close 😄 I especially have problems with the backgrounds...

[onyx thistle]

did you render it in 512 and then upscaled it later or did you render higher?

I also have problems with DWpose, it doesn't really capture the poses right, maybe my source videos are too complicated idk

[hoary kiln]

first pass 512, then 1024, then 2048. Youtube shorts compressed it to 1300+ sadly

[onyx thistle]

and how did you upscale?

[errant geode]

Привет

@onyx thistle привет

[onyx thistle]

привет

[dreamy frigate]

Spent quite some time trying to fix the hands and fingers on this one 😅
https://www.instagram.com/p/C25XvV6MYhh/

ai_guitarist

🎸🏜️ 'Shape of My Heart' - AI flamenco master interpretation amidst a desert cave's serenity. Where the echoes of strings meet the whisper of waterfalls. 🌵💧A blend of technology, art, and nature.

#shapeofmyheart #sting #guitarcover #ai #guitarist #aiart #aiartcommunity #oasis #desertmusic #stablediffision #aianimation #animatediff #acousticguita...

Instagram

[main cliff]

Hi everyone, I'm trying to figure out sd webui, but I can't solve this problem for several days now
Does any of you know the solution?

[lusty vigil]

how much time it took you to generate this animation

[wanton leaf]

https://www.youtube.com/shorts/L07nVJ66OPQ

[fallen girder]

https://youtu.be/Cf2IjQC4T4U

[hoary kiln]

That was one of my first tests, and it's basically a small 30-frame loop with different prompts. It took about 20 minutes to render a single variation and 2 hours to upscale and compose into one.

[icy elk]

is there a way to basically click/drag the controlnet stick figures to alter the pose in openpose?

[snow creek]

Can someone modify this workflow so I can add two loras? please 🙂

[wispy stag]

https://www.tiktok.com/t/ZT8vJkSgB/

[void girder]

https://www.youtube.com/watch?v=KCx0UVWEPQw

[neon lance]

Hey, guys. I've just developed a new system. I'd like to read some thoughts on it, if possible ♥ https://www.youtube.com/watch?v=zptPQbTScto

[earnest aurora]

Increased my autism but looks dope man

[vestal nexus]

https://www.youtube.com/watch?v=TJWgCoAfoyI

[austere ferry]

Sneak Peek🌟- I started this as a fun test with stable diffusion, not thinking it’d turn into a spec ad. I Just wanted to see if I could actually use an AI tool in my workflow. Turns out, AI is a new cool tool for us creatives and it’s very useful. It’s been a real game-changer for adding some extra magic to this piece. Can’t wait to show you what we’ve made, along with a behind-the-scenes look!

Production GOLDCURTAIN @goldcurtaincreative
Supported by @juice.tokyo

Creative/Direction :: @yvesdalbiez
Copywriting :: Michael Glenn @yogatreetokyo
Music: Music by :: nonomi, K.I.
Rap :: Jua
Music direction & production :: @black_cat_white_cat_music
@emanreiff @timoronchino
Sound Design :: Nicolas Amblard @nico.amblard
Cinematography :: Y&N @niccikeller
VFX :: @yvesdalbiez
Dancer :: Hiroko Hizumi @pinko_hi
Skateboarder :: Oba takuma @tetra4298
Breakdancer :: Julien Jolivet @jujujp21
Special Thanks:: Michal Dwojak-Hara @pan_mewa and @juice.tokyo
Nicolas Leon Flores @nicolasleonpantoufle, Alex @osmoze_art, Valli, Jos @mrboofyy, Rémy @remy_creative, Raf @rafpointzero

#MindbodyNoLimits #CreativeRevolution #AIInAdvertising #ai #aiart #aianimation #specad #stablediffusion #stabilityai

[fallow surge]

No

[regal sage]

Would you share the workflow? 🙏

[oblique pecan]

https://deforum.studio/

[snow creek]

do you know any interpolation tool ?

I know there is ebsynth but I don't know how to use it for that

I don't want it to use it for restylization, only for interpolation

[royal zenith]

https://youtu.be/9Z406wtJ1v8

[wispy stag]

💖Follow me, thanks💖 https://www.tiktok.com/t/ZT8cYHnfv/

[hoary kiln]

https://www.youtube.com/shorts/dQzMH8qSRVk

[fallen girder]

https://youtu.be/OFsnOtZFpGE

[plush sandal]

Hi guys, anyone can help with this animatediff issue? I tried to fix it as they suggested but didn't worked for me: https://www.reddit.com/r/StableDiffusion/comments/1ak18iq/cant_fix_error_automatic1111_controlnet/

[neon lance]

how much time is left? - https://www.instagram.com/p/C3IkIe_Oerg/

uisato_

how much time is left?

music by @pulgas_itchy @flanafi
visuals by myself

#touchdesigner #audioreactive #audiovisual
@creativecodeart

Likes

140

Instagram

[noble bison]

https://youtu.be/RTWZSxM7Skk?si=MwWBWyaZkxLolaIq

[wispy stag]

https://www.tiktok.com/t/ZT83L5hyV/

[gaunt parrot]

My frist animation - i mean...not exacly, because i downloaded ready workflow and puts my prompts, my prompts are: red cow with big horns, dog jump, male dynamic pose, and thats it but its kinda cool 😄

[rugged willow]

right on, ya its cool, my first animation was pretty bad lol choppy nightmare fuel mess

[wispy stag]

https://www.tiktok.com/t/ZT83yT4aW/

[dull dune]

#💬｜general-chat

[snow creek]

I don't know why I am getting these kind of vid2vid generations...

does someone have a sdxl beta workflow. or a 1.5 that gives good quality?

[wintry cypress]

lookin good 😮 can you share your learning source material

[fallen girder]

"Rad Storm" a short AI film made with SDXL and SVD v1.1 image2video
https://youtu.be/7imyCXUeTsU

[carmine plume]

Looking for tips on making higher res animatediff's. Just started trying to make videos, but they look low quality and use realistic models.

[livid wasp]

Ancient freestyler based on @indicowie video ⚽️

https://youtube.com/shorts/qonW3jIGjpk

[snow creek]

I've just done my first upscale video

using vid2vid animate diff for 1.5

using an lcm model

[snow creek]

how can I avoid interpolation between first and last frame? it seems every animation is trying to morph them. I am using comfyUI aimatediff vid2vid

[normal copper]

#🎥｜animation

[austere ferry]

Our 50 sec new shortfilm MINDBODY “NO LIMITS” mixing live action and Stable diffusion is live. Watch it now:
https://vimeo.com/913114361

[regal garnet]

Sofa, (minimalist style) , leather, ergonomics, softness, medieval style, fashion, warm color, light khaki color, down texture, Ultra HD, top designer design, unique shape, creativity, powerful design, smooth lines, high resolution images, 16K,

[dusty egret]

This is amazing!!!! watched it in 3 times in a row. congrats! 🔥

[uneven ore]

You won't get much interaction without writing about your workflow.

[fallen girder]

Deforum hybrid motion + Framesync + qrcode_monster, SDXL
https://youtu.be/tTMJbyyEEd4

[pallid valve]

do you guys use runwayml at all?

[strange moth]

Motion Lora Training, Txt2Vid + IPAdapter, colorimetry in Davinci

[wispy stag]

Sora text2video model for stablediffusion when? It produces 60sec photo realistical videos that look insane lol

[candid sapphire]

Anyone familar with coloring the animation?

[verbal sun]

https://youtu.be/IOpjpUOdkK8

[faint rivet]

https://youtu.be/JI3LGJF30pk

[faint rivet]

https://www.youtube.com/watch?v=6Pe9EkuFF68

[scenic kiln]

Made 10 months ago with just frame editing batch style in Stable Diffusion 1,5
https://youtu.be/4NTweEOYg5A?si=HMd_QIsCIgP8W_9E

[sour stag]

sky jump from 10000meters

[gusty seal]

https://youtu.be/DgDLzGYc3qw

[snow creek]

Can someone give me a simple deforum setting to do a zoom in kind of animation starting from an image? please

How can I make a constant zoom in?

[dreamy frigate]

Howl's Moving Castle Guitar Cover vid2vid 🧙‍♂️ 🏰
YouTube (upscaled, was quite tricky!) - https://youtu.be/CBLgKBi508M
Instagram (1024x576, ~1hr to process all) - https://www.instagram.com/p/C3YNpV-MdIz/

Let me know what you think! Always looking to improve for next video 🙂
Also if you like, consider messaging me and following/subscribing, especially if you are a musician as well 🤘

ai_guitarist

🎸✨ Drifting into the magical world of Howl's Moving Castle with a laid-back guitar session. Meet me where magic and music intertwine. 🏰🌌

@joehisaishi.official

#howlsmovingcastle #studioghibli #guitarcover #ai #guitarist #aiart #openai #sora #pikalabs #aiartcommunity #animation #aianimation #acousticguitar #joehisaishi #howlpendragon #animeguit...

Instagram

[snow creek]

My image to video deforum animation looks every image more and more illustrated style and the original photorealistic styles is gone...

can someone help me on this?

like.. I start with left and I get right at frame 20 or so

its like it has worse quality every frame

[rugged willow]

plain prompting t2v AnimatediffLCM

[snow creek]

how did you make LCM look so good? I got terrible results with it

can you share me your workflow

@rugged willow

[rugged willow]

was a workflow i saw on reddit i just added interpolation

[rugged willow]

AnimateLCM with IPAdapter

[faint rivet]

https://youtu.be/d3bRM5-7h6U

[strange moth]

lips-syncing

[rugged willow]

nice, that wouldnt happen to be the talking head model that got pulled?

[livid wasp]

👽 DUALITY 👸🏼
https://youtube.com/shorts/FP7x5myncYo

[novel anchor]

I am currently trying to use animatediff + controlnet from here https://github.com/s9roll7/animatediff-cli-prompt-travel but all my outputs look like this, and the frames are not smooth at all anyone have experience with this?

[snow creek]

If I select cadence 3, so that it only diffuses 1 ni 3 images in Deforum, does it mean that when selecting say FILM interpolation tool it creates the remaining frames? and if so, what happens with FILM if I select 1 cadence?

[white vault]

https://maglit.me/aiacid

[tawdry ermine]

this is really fun to watch! are you using animation for this? or is it a video-to-video thing?

[fallen girder]

https://youtu.be/gU64SPk91FI

[snow creek]

anyone knows how can I use a controlnet video in deforum without being affected by my camera zoom?

It generates zoom because I want it but it applies the zoom to my controlnet video as well which I don't want.

I want my controlnet to stay fixed

[half fog]

hello there. does anyone know how to make a looping animation in deforum?

I was thinking I can use the first frame as a control net and keyframe it to the last frame but havent been able to get this to work.

also forgive me if im in the wrong channel I am new here haha

[void girder]

https://www.youtube.com/watch?v=-OkjBB1WJTY

[lament steeple]

https://tenor.com/bqs7z.gif

[gusty seal]

https://youtu.be/hBIk1koQZZ8

[jovial ledge]

perfect ! how to do that?

[hybrid lily]

https://youtu.be/ff7RiMhB3sw?si=MJ8uN6gDJAw4PN60

[pseudo estuary]

What's this Turbodong...

[lavish basalt]

https://cdn.discordapp.com/attachments/1025266140445933648/1210534653912686592/IMG_0660.jpg?ex=65eae940&is=65d87440&hm=c9ba9995a9538c7934234b9af953d0b9c563fc5d95596474840e98e6b95c0358& Chinese Red Dragon and Little Girl are Talking ¯_(ツ)_/¯

[stable hawk]

AI music, AI voice tools, and etc. Music made with Riffusion inside of Audacity then tweaked.

[tribal needle]

deep forrest in june

[vestal nexus]

https://www.youtube.com/watch?v=PNcxF3SJl2E

[neon lance]

Hey, guys! New system here, what do you think? Would love some constructive thoughts on it - https://www.youtube.com/watch?v=xfF69OTaJi4

[white gate]

im having problems with animatediff

[white gate]

it nearly completes then throws an error, these are in reverse order

C:\actions-runner\_work\pytorch\pytorch\builder\windows\pytorch\aten\src\ATen\native\cuda\IndexKernel.cu:92: block: [150,0,0], thread: [16,0,0] Assertionindex >= -sizes[i] && index < sizes[i] && "index out of bounds"failed. C:\actions-runner\_work\pytorch\pytorch\builder\windows\pytorch\aten\src\ATen\native\cuda\IndexKernel.cu return torch.cuda.cudart().cudaMemGetInfo(device) :92: block: [150,0,0], thread: [17,0,0RuntimeError: CUDA error: device-side assert triggered CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1. Compile withTORCH_USE_CUDA_DSAto enable device-side assertions. ] Assertionindex >= -sizes[i] && index < sizes[i] && "index out of bounds"
failed.

getting this a bunch

could i delete the animatediff extension folder and reinstall it

[white gate]

wondering if this is the issue/solution

[sand holly]

are there any better stable diffusion "how to use videos" I've just been using sebastian kamph's

[humble crypt]

Here's a short animated movie called "Hana", made by using SVD (and a bit of Pika), AnimateDiff with ComfyUI. I hope you'll like it 🙂
(Not sure I'm posting in the right topic?)

https://www.youtube.com/watch?v=FFbTTrIWE5U

[white vault]

https://youtu.be/I8BSIcz1Baw

[white gate]

why does this happen when doing img2img animatediff

should i just use deforum and not use animatediff

was hoping for something simple but this is soo slow and memory exenspive

[rugged willow]

there is a newer animatediff way worth a look, AnimateLCM

[sand holly]

anyone use temporal kit? I'm following the tutorials I'm finding on youtube and it's encountering an error during the first run during pre-processing

[shell flare]

@tame crane

Good riddance.

[shell flare]

can you link me up to it plz?

[rugged willow]

i started with this ComfyUI implementation, workflows are included https://github.com/dezi-ai/ComfyUI-AnimateLCM?tab=readme-ov-file

[safe oriole]

https://openart.ai/workflows/dunker_unknown_58/animatelcm_prompt-travel-fast-video-generation/9iU7dq07Mu503fDSGLin

[white gate]

do u all mostly use deforum or animatediff

[fossil stone]

how do you get such high definition images mine always turns out blurry

[snow terrace]

Deforum, animatediff and svd https://vm.tiktok.com/ZMMdTrGyU/

[sand holly]

have some questions about stable diffusion frames. Would this be where I ask them?

[rugged willow]

i don't know really still figuring things out, played with setting. i try different context option nodes, and i use a different LCM sampler "sampler LCM alternative"

[fair whale]

Is that good?

[void girder]

https://www.youtube.com/watch?v=uFuE-JF8Y64

[faint rivet]

Forget if I shared this... https://www.youtube.com/watch?v=y3W7JzQSdJw

[faint rivet]

https://www.youtube.com/watch?v=UJjfrDFxI04

[oak dirge]

mj + ps + svd + mpc + ultrasharp + premiere + topaz - If anyone wants the workflow, let me know 😉

[nova oasis]

trying animatediff in comfyui, try to talk with me in next step

[nova oasis]

Reference image

[nova oasis]

Btw how to more stable, any library or custom node?

[faint rivet]

https://www.youtube.com/watch?v=6AVoMg0081k

[fallen girder]

https://youtu.be/nWiHs5QnpIU

[strange moth]

Another lips syncing

[shell flare]

@olive halo good riddance!

[clever schooner]

Hey guys. Please could someone give me some tips to make an animation that stays true to an input image for at least 30 seconds. I keep getting results where the image slowly turns into a static colour with no detail anywhere

That's an example of the first and last frame

[clever schooner]

Edit: I think I've found the solution - I had a very low cadence, causing it to change loads of stuff rather than stay the same

[clever schooner]

Edit 2: This actually made it worse. Help still appreciated. As the animation goes on it loses loads of quality - gets worse and worse

[wispy stag]

https://www.tiktok.com/@yooartificial/live?enter_from_merge=share&enter_method=share_copy_link

Currently live, answering any ComfyUi animation questions you have in chat!

[glacial beacon]

This is stunning! What was your process like making this? Did you use thumbnails for scenes you wanted then stitched them together?

[faint rivet]

https://youtu.be/AmUpyju9F7Y

[strange moth]

Skeleton Creative exploration (Touch Designer audio reactive 3d particles to Vid2Vid AnimateDiff LCM)

[clever schooner]

Guys which control net model should I be using? I'm using the depth one at the moment and It's giving me some weird shit

[fallen girder]

https://youtu.be/D9apWuv6aZ4

[round jay]

THIS PERSON IS AI GENERATED AND DOES NOT EXIST!

This AI can make single image sing, talk, and rap from any audio file expressively!

• EMO: Emote Portrait Alive by Alibaba
• https://humanaigc.github.io/emote-portrait-alive/

Reference

• OpenAI Sora in Action: Tokyo Walk
• https://www.youtube.com/watch?v=tRSdt5kmeW0

Looks like people will be replaceable, and its happening very soon 😛

[plain skiff]

This AI can make single image sing, talk, and rap from any audio file expressively!
EMO: Emote Portrait Alive by Alibaba
https://humanaigc.github.io/emote-portrait-alive/

Reference
OpenAI Sora in Action: Tokyo Walk
https://www.youtube.com/watch?v=tRSdt5kmeW0

[wispy stag]

we know lol this is old news now

SD sora when?

[cloud thicket]

Un pasillo de hospital haciendo zoom in

[orchid forum]

What is the best way to do img -> video right now? with a chosen initial image, rather than a prompt. It seems like that is still in progress for animatediff?

[fossil stone]

how did you get the lips to sync that well

[toxic cargo]

||girl||

[idle pivot]

Colab links for animate diff and wwarpfusion?

[white vault]

https://vxtwitter.com/contrastruction/status/1763087550175158502

[stray tree]

https://www.youtube.com/watch?v=xg-JVrHmo3s

[stable hawk]

Original is 3D animated and rendered vid, 2nd is ComfyUI animateDiff vid2vid workflow for a realistic look, mov2mov in Automatic1111 last for an drawn style. Voice made with Bark, lipsync corrected with wave2lip.

[fossil stone]

Do you guys use multiple controlnets for vid2vid?

testing them right now and so far canny .5 openpose 1 has been giving me good results

[stable hawk]

Depends on what I'm doing. I don't use cany much but depth and openpose I use a lot, soft edges I use a lot as well, IP adapter I use situationally. QR Code monster is another ill use situationally for video.

[fossil stone]

What does ip adapter do? Haven't tried that yet, just been testing depth canny and open pose

Also at the end of animatediff, it proprocesses the controlnet again and takes another 5-10 minutes, is that supposed to happen or a bug?

[meager basin]

https://tenor.com/tKgaYjwJq16.gif

/girl

https://tenor.com/dc26eBPYQ6c.gif

[stable hawk]

Takes a lot of the image and puts it into the Gen, style and color palette, a lot of influence from the original with IP adapter. I use it in a lot of my video work flows.

Made this for a group for kids with cancer. They ring a bell when they finish treatment and get to leave. So I tried to include that. Voice made with Bark, music with Riffusion and MusicGen, images made with Stable Diffusion, animated with SVD, and video editing by me. This was a 1 hour project for a daily challenge. I went a little overboard 😅, but still it's only an hour project.

[fossil stone]

Do you have a setting for non img2vid or vid2vid? Just staright up txt2vid usually makes things generate sporadically and don't connect to previous frames

[stable hawk]

For text to vid, I don't actually know 😅. I do freehand illustrations and I do 3D modeling. I can whip a scene up pretty quick these days and bang out a render or a animation to use as an int. With openpose and some of the comfyUI workflows with IP adapter and some others I can use stick figures with a nice render and get consistent video 😅.

[sand holly]

made my first animation:

https://youtube.com/shorts/Kv3ALDU3lEM

If anyone has any ideas how I can improve I'd love to hear them. Especially how to get the character to get their hands infront of the face. I've told it's not possible without editing (i.e. photoshopping), just want to confirm or see if there are any alternatives without photoshopping

[fossil stone]

Animatediff keeps looping the Gifs, is there a way to stop that? Tried using R-P R+P A and N all keep looping

[fallen girder]

https://youtu.be/Ux7FF8zszyY

[white vault]

https://youtu.be/wK4V7qZTm3g

[hybrid iron]

What is vid2vid? And what does it do?

[mental cobalt]

video generator from a video, answer 2: video!

[stable hawk]

Does this

[stable hawk]

I like to make 3D animated videos, run them through a vid2vid to change/enhance/edit them. Can take it any style I want, I can control the background this way as well as it's easy to remove background when you leave it flat with only the character having a depth map and etc. You can technically animate a character, run it though vid to vid, remove background and use it as an overlay animation so you can do multiple characters, it's like an old hand drawn animation technique where the animation are drawn frame by frame on a transparent layer, then they are layered onto the background on top of each other and shot frame by frame to make the animation.

You could technically automate that process with vid2vid and batch remove background then layer them in an editor.

[stable hawk]

Vid for daily challenge in a group I'm in. Challenge was for images favorite game character. But I wanted to make Zelda sing so I used it as an excuse lol. Voice is mine modified with AI, vocal arrangement of the song is the way I sing it to my son 😅. Music is from MusicGen. I took and made audio in Riffusion that almost fit, paired that with the vocals and ran it through music Gen until I found something quirky enough to fit the vocals and sound okay. Video is made from still images animated with SVD then ran through, mov2mov, animateDiff vid2vid workflow, then wave2lip.

[hybrid iron]

How is Stable Video Diffusion different from this?
Is Stable Video Diffusion better or worse?

[stable hawk]

SVD is a lot more sensitive when you try to push past the frame limit. So it's better for short clips. You take those and stitch them together, use AnimateDiff with some controlnets, ipadapters, reactor/face replacer of choice, and chain some other tools to get a much more coherent video from the stitched together SVD animations.

Movie2movie helps a bit because if you set the seed it tries to continue from the last and generates something more coherent. So using that as a base for animateDiff is even better, but not by enough to warrant the effort I out into it lol.

Essentially SVD short gif length clips. AnimateDiff vid2vid much longer coherent video potential with more control.

[noble bison]

https://youtu.be/2PAMRZ9e4FE?si=vhQsaVB6V8ybwbBv

[fossil stone]

sadtalker/wav2lip can take gif inputs?

[stable hawk]

Hmm, can always convert format, I don't think they take gif l.

[vital yacht]

Created with SDXL + Deforum (WebUI A1111 Extension)

100% Deforum (3D mode) diffusion. No interpolation/deflickering, No Controlnet, No Video Init.
Audio: "Pioneers" by Track Tribe (YT Audio Library)

https://youtu.be/pxFO3wUBFaU?si=R-5aAW-vtLzl5FLI

[dark sage]

For all you Deforum users ❤️ https://github.com/Rakile/DeforumationQT

[stable hawk]

Work on forge version as well or just automatic1111?

[dark sage]

For now auto1111, will most likely add support for forge soon. @stable hawk

[stable hawk]

I'll check out, might port it to forge myself 😅.

[dark sage]

Good idea! Should not be to hard, don't think the deforum version differs that much. We add the forge version in the git if you fix it. 👍

[faint rivet]

I have achieved Lego supremacy internally.
https://www.youtube.com/watch?v=uiOdDX5Pmvg