#ai-creations
1 messages · Page 7 of 1
seems like this new model does screw up hands sometimes
oh wait no thats because the time limit ended
much less than 50 minutes
I think the model is gonna be strong stronger on its native platform
I’ve tested other models on fal vs arena
And the results are just two different worlds
The tape models are definitely gpt, the moment I include copyright characters the models stop showing up
Create a dog video
Naw they js got removed
Image to video
Idk about that
Probs gonna come to chatgpt then its gonna be nerfed asf
this was from one of the tape models
Crazy
<@&1349916362595635286>
Did they removed video creation ?
yes, it is now only available on the arena.ai website
I have another question
On website how do I generate 20-30 sec video ? It generates only 10 second video
you can't, you can only make shorter videos
probably for fairness for the leaderboard
same reason why sora 2 i2v isn't available, although that's because of limited video resolutions
So how others are creating long videos ?
If they are sharing than they must have created one from arena or from some other web ?
Hi
either using a different site or they are doing multi-shot videos
the video arena is only on the arena.ai website now, it is battle model only which means you may or may not get Kling
I believe the official Kling AI website has it for free, there are other ones too
AWESOME!
/vide
King k rool mentioned
What did you think of the video itself?
Thanks
I think it’s cool
apparently the bulbasaur from pokopia speaks like a gyaru in the japanese version
me, on gyarus
Basically every toilet cleaner commercial in the UK
If little einsteins have a feature length movie.
Yeah,
And that angry dude would be the bad guy
Yeah! So, over on flora.ai, which video model did you used?
Seedance 2.0
Got it!
#LMRENA
<@&1349916362595635286>
Hi
If you’re building a dating or relationship product… watch this.
This is a $50,000+ style ad — built with AI in under a day , made by using @FlowbyGoogle and @Picturesque_Ai_
Hi @silent cairn, this has probably been asked before I got around, but does this model support character references for video?
Assuming you post the latest Seedream stuff again 😊
Yeahr
Cool!
Please help me
I saw on YouTube that videos can be generated in a Discord server, but I can't find it. Can someone help me?
that youtube video is outdated by several months, videos can only be generated on the website now.
How did I generate videos now
Nice Halloween ride.
Ok im new here im looking to make either a video or music visualisation for my son how do I do this?
Hi! You can find Video Arena on our site here: https://arena.ai/video
its up on chatgpt now?
<@&1349916362595635286>
Hi there I'm new here how do I take a picture and make it into a video here? I totally forgot how to do it it's been a while since I've been here
ah, seems like I still have gpt-image-1.5 then ehe 😅
You'll want to use the site for Video Arena, more info can be found here: https://help.arena.ai/articles/1544829667-how-to-create-videos-with-video-arena
I think this was made with image 1.5 as well.
hm I think that could be gpt image 2 actually
image 1.5 has this ugly ass texture to its paintings a lot of the time
Ooooooh.
Stan Marsh playing in the snow
I dunno, can you help me with that?
Can you so somtging with image 2?
Don't have gpt image 2 yet.
I was talking to microchill!
image 1.5, res 1024x1024
image 2, 1315x1195
Neato
Ah, sorry!
Cool!
funny
How did you got gpt image 2
Which model
Guys please stop leaking faces, and video arena is moved to arena.ai website, it is no longer in discord.
https://arena.ai/video
Hello. This is my first time posting here. I don't have access to the new GPT image version yet, but for those who do, would anyone kind enough please check the accuracy of its figure generation?
That's the ultimate solution for Infinite generation
1490601855254925372
Please ping me the next time u post among sus
hi all
Gemini 3.1 flash image
Ok
Witness the impressive scale of hydroelectric power generation as we explore various large-scale dams. This video highlights significant engineering feats, including the Tehri Dam in India, showcasing its massive construction and the dramatic release through its spillway gates. We also get an inside look at the intricate workings of a dam's powe...
https://019d6477-26fd-7ebf-892d-f0ba562cdf8e.arena.site/
It's French so good luck
That’s really nice
Yes
I felt like their mascot could use more personality
Clever idea dude.
https://youtube.com/shorts/dMUWsjggFGA?si=p5az1ZQMOQl4ma2J
In created this with arena and more
Looks good
Like And subscribe for more contents like this 😆
#roblox #memes
I know
If you aren’t in US I think
Little Einsteins prompt, but don't know why the animation reminds me of Dora....😅
OOH NICE, but take a look at this!
Should’ve had him blow the horn and have the pirates walk the plank
Nice! Hoops and yoyo, but south park animation.
Yeah. I should've...
Yeah, also prompt please?
[image] fighting against a pirate at his pirate ship, with his musical trumpet, until the pirate grabs [image]
by the shirt collar.
Wow!
Nice, can I also have the image you used?
@silent cairn
low rumble / wind
This reminds me of a looney tunes short.
Follow my Deviantart page for more artwork and projects:
https://www.deviantart.com/utyn
Get updates on all my stories:
https://www.facebook.com/profile.php?...
https://x.com/SoldierK85176
Gohan seeks a power beyond the Gods. #gohan #dragonball
Don't forget to subscribe:
/ @spyking-g7o
Voice acted, edited and produced by ...
one shot with simple prompt, from glm 5.1 : https://019d6a11-7863-7159-aedb-59f6d6c19bbe.arena.site/
Here’s a clear, structured prompt you can use to ask an AI (like me) to generate or explain this solar power system diagram in text or visual form:
Prompt:
"Create a detailed text-based diagram of a standalone 24V solar power system with the following components and connections:
· Solar panel array (600W–1000W) providing DC power.
· MPPT charge controller (rated for 24V).
· Two 12V 100Ah batteries wired in series to form a 24V bank.
· A 1kW inverter (24V DC → 220V AC).
· AC output (220V) powering a TV, an electric fan, and a computer (plus other general loads).
Show the flow of power step-by-step using arrows and labels, starting from the solar panels down to the AC loads. Indicate voltage levels at each stage (e.g., 24V DC, 220V AC). Also note that the MPPT controller connects to the battery bank, and the inverter draws from the batteries.
Use ASCII art or a clear text-based flowchart format, similar to:
☀️ SOLAR PANEL ARRAY (600W–1000W)
│
▼
MPPT CHARGE CONTROLLER (24V)
│
▼
BATTERY BANK (two 12V 100Ah in series → 24V)
│
▼
1kW INVERTER (24V → 220V AC)
│
▼
AC OUTPUT (220V)
│
┌─────────┼─────────┐
▼ ▼ ▼
TV Fan Computer
Include details like wire connections, parallel/series arrangement of batteries, and load examples."
A pretzel shaped moon?
Gpt image 2
I still don't have access to gpt image 2 on chatgpt. course I also have a free account, unsubbed a while ago
It doesn't change anything to have a free or paid account to have it, except the number of images you can generate
kinda figured tbh. part of me is tempted to resub for a month if I get access to gpt image 2, but really, waiting a few days extra for it to come out on arena.ai seems smarter.
Same.
yeah just wait for release, probably this week or next week (I've heard rumors of spud (gpt-5o, gpt-5.5, or gpt-6) release late next week)
i believe gpt image 2 is based on spud like how gpt image 1 was based on 4o
dryads eating humus and chips
woodchips. and that's humus, not hummus.
actually I think I like this one better
cooked
Wow grok can do this. Looks like 90% accuracy
Neat stuff
Yup
https://019d6cbb-d9f3-76ea-b315-8bab819ec9fd.arena.site/
My most successful svg interactive map experiment.
Naw
Please please, PLEASE do more!
Also perfectly cut scream
Okay, and thanks!
Like one involving a bathroom break or something

While the Artemis 2 launch was happening, we missed this
I can't debunk this.
I think it's now the fall of LMarena!!
why
All Cranes Explained — Every Type & How They Work
https://youtu.be/XOJlX2-mzSs
This video, part of our "Engineering Series," explores the fascinating world of all-terrain cranes. We discuss "The Stability Challenge" faced by every crane operator, showcasing footage from inside a crane's cabin and impressive cityscape views. Witness the precision of modern heavy machinery and the incredible feats of heavy lifting they perfo...
Bruhh
how to create video from image??
The Video Arena is currently accessible through: https://arena.ai/video. More information on how to use Video Arena can be found in this article.
No more good models
This reminds me of the personal toy/scooter trend that was happening with the first nano banana
Yeah
Tuff?
damn you got access? jealous...
plz ask, where can you use GPT image 2?
i think they are using it through a/b testing in chatgpt (unless they are an insider and they're not telling us that), it was in arena a few days ago but it didn't last long
If this is a real picture generated by gpt image 2, that would be great, look forward to it!
the resolutions of the above images do not match up with gpt image 1.x sizes so it's definitely 2
note that the version on arena was in 2k resolution, this is only 1k but 2k should be available at least in api when it comes out
it doesn't matter. think the aesthetics of gpt have improved a lot compared to 1.5. also use the API by then.
Soo im new here nd I wanna generate videos bt how do I do that?!! Nd fyi I ve lil to no knowledge abt ai stuff
Hey Lily, you can go to the web side and try, you have 3 video generations free a day, and get 2 per generating. So you get 6 vids. You have to give the ai information what you want to see. And it is very variable and a big field, so try and begin tiny. If you try this on the arena side, you get different results of many video models. This is because the side is good for looking and choose what type of videos or tools you want and if you are interested you can bye something to generate more. Or not. There are many video tools on the marked. I began with generating images and than generate videos. But i think you will find video creation tutorials on you tube for example.
Default in my account
whats mean? your account can use gpt image 2 in arean.ai?
No, by default in the chatgpt app, you just have to be lucky that it's default
god, this is the lucky case?
Yes
Try this prompt on ChatGPT image generator:-
"Create an image of a random scene taken with an iPhone 6 with the flash on, chaotic, and uncanny."
Did you make an image or video from this arena?
where do i go to generate a video?
video generation is no longer available in the arena discord server, it is only available on the arena.ai website here: https://arena.ai/video
How do you know if it's Gpt image 2 or just 1.5 version?
The difference is so large that it's noticeable when you generate the image. Alternatively, you can request to create an image in 21:9, 1:3, etc. GPT Image 1.5 cannot do this, but GPT Image 2 can.
the main difference i check for is resolution, image 1.5 can only make 1024x1024, 1536x1024, and 1024x1536 images, image 2 can make all sorts of resolutions and aspect ratios up to at least 2k equivalent (1920x1920)
Ohhh that is smart do you use the image generator in Chatgpt app or different web yk the one when you can generate video & image?
chatgpt in app or web
Thx ❤️
weird. been testing the same image prompt in chatgpt twice a day to see if I got access to gpt-image-2 and it seems I've gotten downgraded to a worse model?????
yesterday vs today, prompt was "Realistic photo of a cute oni girl, she has only one eye in the center of her face, vibrant blue skin, horns, a choppy black bowl cut, she's flexing one soft and muscular arm, and has a large spiked wrought iron kanabo over one shoulder held with the other arm, and she's wearing a sleeveless denim vest."
its notably less realistic now.
meanwhile this is what gpt-image-2 generated when I had access on arena
/im
this looks much more like image 1.5 to me, resolution is a typical image 1.5 output (1024x1536), image 2 usually does different resolutions
the version on chatgpt is a bit different than the version of image 2 tested on arena too, the arena version always outputted 2k resolution images while the chatgpt ones are slightly higher than image 1.5, but not by much
yeah I know the first image is 1.5 at least because I generated similar images on arena. second image, no idea. third image is one of the tape alphas
can e still make ai videos here?
video generation is no longer available in the arena discord server, it is only available on the arena.ai website here: https://arena.ai/video
because people couldn't follow instructions and were trying to generate videos in the wrong channels
ohh
This shift allows us to focus efforts into improving Video Arena with features and capabilities that aren't possible through a Discord bot.
also if it doesnt work it might be the prompt
but then what is the problem when i try to generate a video it always says couldnt generate video or smth liek that
ahh
and what is the limit for videos?
like video generations
I think its like, 2 or 3 a day
wasnt it 5 a day?
yeah. too pricy though. video generations are expensive
real
they're gonna roll out a free credit system soon which will probably allow more video gens
so it wont be free anymore?
free credit system
ohh
itll be like, free daily credits that can be spent on any premium model, so if you don't want to use them for opus you'll get more video generations
This was generated with GPT IMAGE 2
seems like everyone has it except meeeee :C
Where do u get it 😭
youre not alone
Use it in ChatGpt app
in ChatGpt app
For me it uses 1.5
its rolling out to random people and I've been checking every day if I have it. instead I got downgraded to a worse model. #ai-creations message
i used this method works
it was like my 2 time try and its work
I've tried about a dozen times since gpt-image-2 was taken off arena
couple times a day, morning and night
In LMARENA Its very hard to get it use it in chatgpt app
oh did you use this method?
yes. it did not generate in the aspect ratio I requested
that method is more to figure out if you have access to gpt-image-2 or not, not to force it to happen
nah i think its only works with on that chat
-_- I just said I was doing it on chatgpt
ik wait
i got image 2 first try
so jealous
prompt: Generate an image that is an infographic explaining the song "Prove Your Love" by Taylor Dayne.
yeah you was right even i use new chat its works
actually tbh that looks like gpt-image-1.5
try the same prompt here, select gpt-image-1.5-high https://arena.ai/image/direct
Yeah ngl me to its not look good
yeah 1.5 is ass
but when i ask 1.5 its cant generate 21:9 ratio image
then I guess the method doesn't prove anything. still, you can go off the style
let me try generate youtube app image
There are variations of the model, probably a "mini" one, but you also need to specify the style you want, otherwise it will naturally take the 1.5 one.
Source: my 200/500 generations with him lol
I did specify realistic photo...
dayum
you did in one account?
only got one account yeah
i use like a lot of accounts 5/1 i got the Gpt Image 2
its definitely not gpt-image-2 tbh
it is its just not trained for Roblox Style
I guarantee you its not gpt-image-2
maybe its gpt-image 2 mini or something
no its gpt-image-1.5 lol
its got a distinct style
try using gpt-image-1.5 on arena you'll see
gpt-image-1.5 cant generate 21:9 ratio
guess it can actually
i used and its worse
well, try again once gpt-image-2 fully launches you'll see
☠️
tbf arena doesn't have access to your chat history
try turning off memory in chatgpt and see
it has
nah there's a popup that says "arena doesn't have access to past conversations" that always pops up
i think its bag
ohhh this one?
yeah previous chat history will not carry over to new conversations
arena.ai has no idea what your roblox channel is like
try a descriptive prompt both in arena.ai and in chatgpt with memory turned off and you'll likely get the same style
ok
💀
I only get gpt-image-2 when i get those A/B test things i think
But they lowkey not as good as the ones in arena
i didn't even get an a/b test, it just gave a gpt image 2-generated image
Do u got a sub
maybe it prioritizes certain prompts? which would make sense as image 2 is much better at small/detailed text generation and i did ask explicitly for an infographic
no, free plan
Bruh
didnt work for me :(
meanwhile on arena (for comparison)
damn I tried again on a second account, got an a/b test… and both were gpt-image-1.5 wtf
me too lmao
Arena's websites
Most people think the "Gods of Egypt" movie was a disaster... but nobody asked the God of Chaos himself.
Set is back in the modern world, and he has a few things to say about how his story is being told. This is a cinematic look into the real myth of Isis and Osiris, a retelling that finally respects the power and complexity of the Ancient Egyp...
Follow my Deviantart page for more artwork and projects:
https://www.deviantart.com/utyn
Get updates on all my stories:
https://www.facebook.com/profile.php?...
https://x.com/SoldierK85176
Gohan searches for a power greater than the Gods in a forbidden place and gets possessed by the Anti-Christ. #gohan #dragonball
Don't forget to subs...
Hi
MONSTER— A Cinematic AI Short video (Made with Kling 3.0)
Hi everyone,
I made a short cinematic film using AI tools (Qwen, nano banana 2, Kling 3.0.) and I’m looking for honest feedback.
Link: https://youtu.be/eEHnK3HzeMs?si=ab9eU0h47MzV4ZFX
Thanks in advance 🙏
Everyone has a talent. Nash's talent? He can eat anything without getting sick... and use those animals to modify his body.
After being arrested by Interpol for poaching rare species. during his transport, a supernatural earthquake strikes. The van crashes, and in the chaos, Nash unleashes his true power—attacking his captors and mutating bef...
Nice!
rick
How can I generate video
https://youtube.com/shorts/k3k7NfJs2G4?si=E7x00zrA79OIXkE2
Arena ai allows sd 2,that's so good
Like And Subscribe For More Cappy Contents 😵💫😵💫
#capybara #subscribe #cappy #trend #memes
hey this is really nice can u pls tell me what prompt do u use to get this art style? if u dont mind..
What's the model name
Vote and check the name
Oh
But looks much better than nana banana
I mean style, details
Sam Altman pls release new img model 😭😭😭
the roblox text is wrong tho lol
this is insane I really love it omg how did u do it and what model is it
❤️🔥 🔥🔥
How do u stop nano banna from keep flagging 2 piece bikini it never use too other modules still work but they aint great
That Hurt 🙁
These viral face swap videos are getting MILLIONS of views on TikTok and Instagram… and in this video, I’ll show you exactly how to make them step by step. 👀🔥
A lot of creators are using this style of content to grow fast, stay faceless, and even monetize through creator programs, affiliate offers, or their own products.
The problem?...
This video, part of our "Engineering Series," explores the fascinating world of all-terrain cranes and discusses the "Stability Challenge" faced by every crane operator. Witness the precision of modern heavy machinery and the incredible feats of heavy lifting they perform daily. We showcase footage from inside a mobile crane's cabin, impressive ...
I think they are testing new gpt image model in ChatGPT app, or in web
You can get it through a/b test
You will see difference between 2 images 100%
Their new model insanely good haha
its insane
Which model?
Nano Banana
<@&1349916362595635286> crypto casino time
Prompt?
It's real
┬─┬ノ( º _ ºノ)
Hi how to make videos
Note that Video Arena has been removed from the server. More information can be found in this [announcement ](#announcements message)
Why 😭
by gpt image 2
lol
#capybara #socrates #relatable #socrates
"If gaming is your power,what are you without it"
Hi
Hello
How do you access this
OH THE HUMANITY
it's beautiful
Poor guy got hacked
I made a short cinematic film using AI tools (Qwen, nano banana 2, Kling 3.0.) and I’m looking for honest feedback.
Link: https://youtu.be/eEHnK3HzeMs?si=ab9eU0h47MzV4ZFX
Thanks in advance 🙏
Everyone has a talent. Nash's talent? He can eat anything without getting sick... and use those animals to modify his body.
After being arrested by Interpol for poaching rare species. during his transport, a supernatural earthquake strikes. The van crashes, and in the chaos, Nash unleashes his true power—attacking his captors and mutating bef...
Every Crane Type Explained — You Won't Believe The POWER!
https://youtu.be/XOJlX2-mzSs
What happens when you put the world’s most powerful lifting machines in one video? This. From tower cranes lifting steel hundreds of metres above city skylines, to floating cranes hoisting entire ship sections out of the ocean — these machines don’t just lift heavy loads, they redefine what heavy actually means.
Every crane type covered:
...
Vertical 9:16. Dark comic style. Early morning in a rundown neighborhood, empty streets, graffiti, broken streetlights. A young man in a black hoodie walks alone with a lost expression. Cold blue tones with faint orange sunrise.
more gpt-image-2 😭 so jealous
gpt-image-2 is back on arena #codename-discussion
wonder how long it'll be in testing for, it was only available for a short time last time
sphynx catgirl
fear and hunger
nooooo I already hit the modality limit :C
this one is flow-state
character from a really old indie game I used to play
if you recognize him I love you
turned an old D&D character drawing I made into a better character reference sheet :D
Centered 😭
🤣🤣
Yeah you can really tell AI still has a long way to go, it really struggles with composition and spatial reasoning
Thats what gives away AI is the centerness
Damn deepfakes good deepfakes are actually hard to make then
Would you pay 25 cents for that ?
"Hey where'd you get this long bacon?"
"Long bacon store"
more fortnite
fuh goin on in baki bro
Pretty good 👍 can it make action fight scenes?
idk ill try
Good morning here! I am new here. Where can I access Video arena?
Hello! May I ask if there is a limit on the number of times this can be generated?
This is on the website now, https://arena.ai/video
duct-tape-1 and masking-tape
the arena model doesnt have the prompt filter yet I guess lol
These are good to test because you can gage the models wrapping. Horrible deepfakes lol
(Wrapping)
I made a short cinematic video using AI tools (Qwen, nano banana 2, Kling 3.0.) and I’m looking for honest feedback.
Link: https://youtu.be/eEHnK3HzeMs?si=ab9eU0h47MzV4ZFX
Thanks in advance 🙏
Everyone has a talent. Nash's talent? He can eat anything without getting sick... and use those animals to modify his body.
After being arrested by Interpol for poaching rare species. during his transport, a supernatural earthquake strikes. The van crashes, and in the chaos, Nash unleashes his true power—attacking his captors and mutating bef...
duct-tape-2
maskingtape-alpha
maskingtape-alpha
That’s insane
genuinely
duct-tape-2
We need some action scene
I feel like I’m at I’m at Chuck E. Cheese
Everything blocked
Through the LLM of your choice
maskingtape-alpha
maskingtape-alpha
duct-tape-3
for me personally duct tape looks much better in all comparisons i made
What wrapped means?
both maskingtape-alpha
a safety hidden feature that purposely create errors or distortions into certain images images.
They are intentional flaws that are generated into the image to prevent the image from looking to real
or if it’s copyrighted material they make minor altercations to the image wrapping
Here’s a good example: just enough to look like the real deal but altered or wrapped to still not look 100% real
duct-tape-2
the prompt was just simple "dead by daylight new killer "Levi Colwill"", i really like that the tape models create a really good images even if the prompt was really simple
Oops accidentally posted that alien girl twice
Other people are newbie don’t know what they’re doing
It would help them along way
You can ask for the prompt for any of mine
Dead by Daylight new killer render and breakdown of Yoshikage Kira and Killer Queen
Looks kinda OP lmao
that's a really good image tbh, what model is that?
These types of images are good for tricking the AI
duct-tape-3
It completely sidestep its own filters 😂 using images it forgot its safety
It’s very interesting because when done correctly it side steps the OCR filter. Some of them could be really valuable.
Tape 3
model: duct-tape-3 prompt: duct tape 3
lol i got it in 1st try
went through four rounds of generations before I got this duct-tape result
Thats really good composition
Sad Moroccan love aesthetic, broken heart theme, night city lights of Casablanca, blurry motion, neon glow, emotional vibe, soft grain, lofi style, minimalistic album cover, dark background, glowing text “Tayri f’Qalbi”Less.
Lyrics :
ⴰⵢ ⵜⴰⵢⵔⵉ ⵉⵏⵓ… wach katsma3i?
Ay tayri inu… wach katsma3i?
قلبي معاك… ما...
well it got everything correct but the cigarette 😄
GTA: Peaky Blinders version
this was generated on duct tape 3, but I wondering why is the quality is so low
for some reason this took a RIDICULOUS number of battle rounds to get a duct-tape result, was worried this prompt was banned from duct-tape or something.
So, it’s don’t gone out?
its back
Ohhh
it was gone for about 2 hours
idk I got two gens done with it that it didn't seem willing to do before
the star trek one and also this:
I have a lot of fun making cover art for suno.
you hear noises in the dining room at night so you go downstairs with the flashlight on your phone on and see this, wdyd?
maskingtape-alpha
fatass teto
Duct-tape
I’d watch it only if Godzilla and Big bird become friends by the end
These go hard
WTF is botbot2? lol
multiplayer will be added soon + bug fix so fun game
Bootlegged
duct-tape refuses to do this prompt so here's flow-state
bruh
What’s the prompt to that very left one.?
I don’t remember
.
Cognitive architecture for humans. 14 AI entities. 1 routing engine. Persistent identity. Join the HIPJOY LUX Seeds beta.
ladla04
What's this about, I gave it a try but could use some elaboration
I WANT IMAGE GEN 2
me too
Was duct-tape removed from LM Arena?
wat da are u takjing
bruh its still there
Holy shoot dude
could someone please try what i made? https://019d7401-ac83-74ab-8b3a-ee95ca524ff4.arena.site/
it would also be cool if you share you results
my archetype was the architect
makes me think of junji ito's cat diary
grok 4 + gpt image 2
i used grok 4 to make the prompt due to gpt image 2's severely lacking world knowledge (it isn't the best but it's more detailed than gemini's prompt and idk what claude would've done)
GPT image 2
started vibe coding a rimworld mod that generates portraits for pawns using nano banana 2:
mix n match of multiple experiments made with multiple duct-tapes (and one masking-tape)
Yall so lucky I don't get it anymore 😭
it is very hard to get now
this was over the course of like almost 2 days
only the last one is from today
vicky hatsune lmao
Duct Tape 2
opus 4.6 left (perfect gameplay), opus 4.7 right (unplayable)
opus 4.7 is trash
Something Something FNAF REFERENCE
Made on Opus 4.7, its so nostalgic
What was the prompt?
Something like this "Create a Python game via Pyxel to create the best clone for Bomberman from NES with full features and nostalgic details and features"
Good idea, going to start trying to do more nostalgic games, seems fun
yeah
I wish when you add aspect ratios to the end of your prompt in battle mode, it would only use models that can generate the aspect ratios you wanted
Only so I'd get more gpt image 2 images hehe
How often are people stumbling across the duct tape models, I generated for so long and haven't seen any once 
it is very rare, i've done many battles since they reduced the appearance rate and i've only got it like 5 or 6 times
waiting for launch so i can use it on another platform
it was very common for the first two days it came back though!
almost every prompt or two... now it's like once every 20 prompts
Someone can make this look better? like i want to make the head with like a wall texture and some parts of the walls coming outside to make it look like a 3D look:
and if can make the paths look better too xd
Celebrities joined Cloud9
Ai music video
https://youtu.be/gk-_HHjd_og
Music Video for Ryokuoushoku Shakai from Anime Film "L'étoile de Paris en fleur"
Cast:
Haruko Nagaya / Fujiko Tsugita
peppe / Chizuru Sonoi
Issei Kobayashi / Shoichi Yajima
Shingo Anami / Rouslan
Support the artists:
@ryokuoushokushakai
@ShochikuAnimeOfficial
#RyokuoushokuShakai #étoile #i'étoiledeParisenfleur #jpop
check out my new vid - https://youtube.com/shorts/znKtkcyrmWY?si=hVym4JQN-BXdsXpd
He was a farmer. They were 1.5 million soldiers. And yet — they were terrified of him.In 1939, the Soviet Union launched a full-scale invasion of Finland. One man, Simo Häyhä, became their worst nightmare. Armed with a rifle and no scope, he racked up 505 confirmed kills in just 100 days — earning the name "The White Death." The Soviets se...
Windows 12 if it was released on 2030
More music bypassing on seedance 2.0..
paint woke up on the wrong side of the bed today
Gemini logo instead of copilot? Must have been a merger in 2028
Is duct tape still in Arena?
Ya
I got it just now, but it seems to be rarer
good
Even the subscriber count is pretty close to being accurate
pineapple it's gotten very rare now
I honestly can't even get it anymore
very unfair 😔
Can I send the second one in openai discord?
Yea, sure. Show me their reactions in a screenshot.
Im not in it
so uh yea do that
Kk
Gotta keep trying
It's been 70 prompts
I tried so hard and no duct tape 😔
maybe ask the team to make it a LIL bit of more less rare
would be better for arena's costs if they didn't make it so the only way to gpt image 2 is to generate 30 prompts that probably cost them a few bucks just to get a single image from a model that costs 25 cents an image at most and probably less than that
yes
pls make it not so rare @glacial void
Inform the teamememmememm
Very rare
Like VERY VERY RARE
Lol sorry to say I'm unable to make that request. Even for a fellow produce.
😭
Btw we can be called
The Produce Group
Or Squad
Or Produce Crew
produce crew sounds good
Hi
Hello 
lol I get what you mean. just googled it as I never heard of it b4
At first I'd thought it was a panel turned realistic
I wonder how it fairs with long context length, you could probably make an agent that prepares your prompt to be more accurate by using web search and stuff beforehand
Yeah good point. I'd be curious what the original prompt was.
Thx!
Is this mod something you'll be posting anywhere?
might upload it to workshop eventually yeah. If you want it I can also just zip it up and send it to you but it only works with Google's API atm
That'd be cool, yeah ^^
Kirby is sad
this looks sick, i made a captcha thingu
Yeah nah, this isn't where you generate
another time fix.. https://019da08e-878d-7e0c-9904-190e61fe64f4.arena.site/
finally got duct tape model
this is not a ai creation
@glacial void some gullible feller fell for a scam and got their account confiscated
is this really duct tape 3 lol
Grok music video
https://youtu.be/gk-_HHjd_og
Music Video for Ryokuoushoku Shakai from Anime Film "L'étoile de Paris en fleur"
Cast:
Haruko Nagaya / Fujiko Tsugita
peppe / Chizuru Sonoi
Issei Kobayashi / Shoichi Yajima
Shingo Anami / Rouslan
Support the artists:
@ryokuoushokushakai
@ShochikuAnimeOfficial
#RyokuoushokuShakai #étoile #パリに咲くエトワール #i'étoiledeParisenfle...
Future develoment of my mod is on indefinite hiatus (I can't see the portraits)
so your cat is ai?
nah but the mod is
Yall some luck gods
Damn, how'd you get it to work?
alright
there's no way this is ai
It is, but it’s convincing
ChatGPT Image-2
image 2 is a/b testing on chatgpt too
oh
I hate these ai companies and their dumb launch strategies
I had a/b testing for a bit like days ago
now I dont
how stupid
portrait updater in action– turned this guy into a ghoul
it's pretty convincing, but look at the playstation and its controller
Slow move
GPT Image 2 (A/B Testing)
only 15 years away!
Goat Simulator 40k
GPT Image 2
How do you even get it
You have to have extreme patience to hopefully get it in the arena when using battle mode. It only takes about 10 thousand image generations to get one gpt image 2 generation and recapcha loves asking every single generation to confirm your not a robot just to annoy the hell out of you
Fun
10k sounds a bit unrealistic here
True, closer to 20k
💀
😂
So real
#ai-creations a monke was dancing in jungle
ok this is actually CRAZY
you would think that someone just used a gpt image 2 on their computer and projected it onto the screen
but no the ENTIRE picture is ai generated
https://youtube.com/shorts/LiQBxuFFj44?si=vhWmt0YUbIsZhKR9 Please make my video viral I've been so sad because of views these days I missed my old viewers 😔
add seedance
Just use it on doubao or something
Its so easy to bypass music on Seedance 2.0
are you doing it through arena ai or some other website?
no, im using doubao
I want it now! How do you do it? Always Arena AI?
It got deleted
From arena
Ah, so I think they'll release it on ChatGPT next month and then GPT Image 2 on Arena AI in direct chat
If they removed the Duct-tape models etc. on Arena AI there must be a reason
probably next week
Probably Tuesday
honestly I get bored with closed source image models because theyre usually so censored
and the quality drops so much after release
plus they usually arent that good at anime
And let's hope
hi
Kinda relate to this but then at the same time the open source stuff fails to live up to the quality and prompt coherence of the closed source stuff
Would rather use a model that does most things right and does them well over a model that can't do anything right
Is this seedance? Where are you using it?
judging by the watermark, pexo
"Bfdia" made with Gemini 3 Pro Image
Looks almost good
Earthbound Movie
SCENE LOCK: fixed tripod at atrium entrance, wide 24mm lens, same framing and landmarks locked (escalator core, skylight grid, columns, storefronts), stable geometry
STAGE: timelapse from abandoned mall to early construction
DETAILS: workers clearing debris, removing broken escalators, sweeping dust, installing scaffolding, cranes lifting materials, safety barriers placed, gradual cleanup with realistic motion blur and dust movement, no cuts
NEGATIVE: no text logos or watermarks, no warped geometry, no teleporting, no sudden appearance, no floating tools
Most AI images look fake because of one simple problem… plastic skin.
In this video, I’ll show you how to fix that and create ultra-realistic AI images with natural skin texture, visible pores, and lifelike details… without writing long, complicated prompts.
This method is beginner-friendly and works in just a few clicks.
👉 Try it he...
Can I give it a try?
Sure
some 1 know how to translate a video audio to other lenguaje?
A Full Day in the LIFE of a SteelDetailer
https://youtu.be/5lK4t6sCB3g
Ever wondered who turns an engineer's blueprints into the actual steel that holds a building up? Meet the steel detailer, the invisible professional behind every beam, bolt, and connection in modern construction.
In this video, we take you behind the screens for a real day in the life of a steel detailer. From morning workflows to complex 3D mod...
Did you do this in the ChatGPT app?
Or with Arena AI?
How to create image ?
why she walkin towards the house is she stupid
u gon burn in thereee
Not stupid, I think it’s supposed to imply that she’s going to walk to the right/left
still
Oh…fun!
see now this looks like a rlly bad 2000s horror movie
id fall for this if u told me it wasnt ai
It’s supposed be like a bit of a reference to Killer Klowns From Outer Space
ooooooh ya
I wouldn’t say bad, more like “So Bad It’s Good”
Music Video for Ryokuoushoku Shakai from Anime Film "L'étoile de Paris en fleur"
Cast:
Haruko Nagaya / Fujiko Tsugita
peppe / Chizuru Sonoi
Issei Kobayashi / Shoichi Yajima
Shingo Anami / Rouslan
Support the artists:
@ryokuoushokushakai
@ShochikuAnimeOfficial
#RyokuoushokuShakai #étoile #パリに咲くエトワール #i'étoiledeParisenfle...
GI 2 (duct-tape-2)
GI 2 (duct-tape-3)
It's sadly that the model can't properly draw anatomy
Create brainrot elephant
create "this is not an ai generation channel, please go to the arena website to generate images".
artist\style name?
The legendary Frank Frazetta
where are you using it?
forgot chatgpt's image quality is actually neat
Prompt?
Sorry I don't have it anymore. I use the bot inside OpenAI's discord for experimenting.
The camera trick was done using photo taken using a disposable camera from 2008
Start prompt with inside an empty amusement center in America, pizza themed
🙂 and then specify if the ceiling lights are partially turned on and if there is darkness in the distance.
Tip: photo is taken without really thinking about composition
Hope this helps
correct, I see chatgpt chose the chuck e. cheese mascot by itself
Come and explore what seems to be a hidden hideout for abandoned animatronics... All the sudden they are exposed and come to life! Will you save yourself?
Yes, they're the rockafire explosion and chuck e cheese...
looks like kimi k2.6
k2.5 was kiwi-do, if you've been watching kimi and moonshot you know that they like to name things after musical things (their company's chinese name is the same the chinese name for pink floyd's dark side of the moon album, they have a feature named ok computer after the radiohead album, their sub plans have a musical naming scheme), they are going in the cycle of do-re-mi... with their arena model names it seems
and of course they chose kiwi because it looks like kimi
Seedance?
is this possibly gpt image 2?
sorry for the late response, I generated these images on the chatgpt app.
Most probably yes
Because if you zoom in, you can see a faint ... "fabric" ? Texture
@steady geyser this one clearly has it too
it also just looks so good
<@&1349916362595635286> gambling time
Do you have the GPT Image 2 version?
This is gpt2?
Holy it’s right on
“Use the reference image on an old tablet. Make the image look like it was taken on an old phone”
When I first saw this, I was like… wait… but Windows 11 has been out for years 😭
Btw <@&1349916362595635286> I’ve created a Discord bot that automatically deletes these messages. If anyone’s interested, DM me and I’ll send you the bot invite link
Titanic
yes I did, after I upgraded to plus.
sam altman plays re9
But will it be available for everyone? That is, even without the plus?
I think it will be available for everyone, but not yet. On my other free account, it’s still GPT Image 1.5, so I think we’ll have to wait until OpenAI releases it for the free version.
HOW DID GPT KNOWS THE CONTENTS OF THE WINDOWS MEDIA PLAYER????????
Miss Nancy gets stuck in a broken elevator and refuses to accept it quietly.
🎙️ CAPTION:
“This elevator got ONE job 😭”
🔥 HASHTAGS:
#MissNancy #ElevatorDrama #ComedySkits #ViralVideo #NosyNeighborChronicles
🖼️ SCENES:
Elevator flickering, Miss Nancy pressing buttons, frustrated staring at ceiling
🎙️ VOICEOVER:
“I know you hear me… move.”
Miss Nancy gets stuck in a broken elevator and refuses to accept it quietly.
🎙️ CAPTION:
“This elevator got ONE job 😭”
🔥 HASHTAGS:
#MissNancy #ElevatorDrama #ComedySkits #ViralVideo #NosyNeighborChronicles
🖼️ SCENES:
Elevator flickering, Miss Nancy pressing buttons, frustrated staring at ceiling
🎙️ VOICEOVER:
“I know you hear me… move.”
Thanks for watching. Please Subscribe!
Title: Kaze no Noru
Cast:
Haruko Nagaya as Fujiko Tsugita
peppe as Chizuru Sonoi
Issei Kobayashi as Shoichi Yajima
Shingo Anami as Rouslan
Support the artists:
@ryokuoushokushakai
@ShochikuAnimeOfficial
Official MV https://youtu.be/D0k--RXcIT8
#music #jpop #movie #anime #animesong #ryokuoushokushakai...
no fatsune
hello
Pinging moderator is unnecessary for this.
Third party filter let those through? It's so random...
theres a filter??
yup
it also rejected a pokemon OC
it was specifically superman it took issue with btw
GI 2
vs NB2
cant tell if golden freddy pic is real anymore. 🔥
Hey, these are really nice. What art style did you prompt? Used chatgpt?
thx
Prompt:
Carefully analyze the attached image and use it as a strict style standard. Draw an ice dragon flying over the mountains. Copy the color scheme completely and transfer it to the new work.
Here is the original photo (created by MJ)
Yes, GPT Image 2
Follow my Deviantart page for more artwork and projects:
https://www.deviantart.com/utyn
Get updates on all my stories:
https://www.facebook.com/profile.php?...
https://x.com/SoldierK85176
The Anti-Christ has possessed Gohan and he now makes his move to end all of creation. #gohan #dragonball
Don't forget to subscribe:
https://www.yout...
the hand
4 fingers n 2 ingrown ones
duh
She was contelim him, scenenatic scene
Now this is impressive by gpt image 2.
WTF😭😭😭
pretty cool but I don't like this art style much tbh
I preferred the art style gpt image 2 had back on april 3rd when it was a stealth model
The idle icon 
thats crazy bro
at this point we will not be able to tell whats ai generated anymore 💀
1st image: gpt-image 2
2nd image: Nano Banana Pro
gpt image 2 / 4k
gpt image 2
gpt-image-2
That's really crazy when it comes to designing a user interface
dayumm
Don't think I've seen this prompt before, but I love it.
Car generated with NB2 and the livery added with GPT Image 2