Does anyone have Sora 2 experience on chat gpt pro vs plus? Is sora 2 worse than Sora 1 in terms of resolution? Is it because I’m on ChatGPT plus. I cannot make a video with usable resolution. And the ones I see all look 480p at best often worse. They all have the watermark. It’s fantastic at following prompts but the end video is terrible. And you can’t even select aspect ratio resolution or the number of videos, I’m really confused by the way it’s rolled out, it’s meant to be cutting edge but it’s presented like a cheap app for Tik tok kids
#Why is sora 2 resolution much worse than Sora 1? Is pro and plus membership different for sora 2 spe
tall rose
outer basaltBOT
Hi @tall rose. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:
Based on the available knowledge sources, here’s what’s clear about Sora 2’s resolution, membership differences, and user experience:
Sora 2 Resolution vs. Sora 1
-
Sora 2 supports higher resolutions than you may be seeing: The official specs for Sora 2 are 1280x720 (landscape) and 720x1280 (portrait) for the standard model, and up to 1792x1024 (landscape) or 1024x1792 (portrait) for Sora 2 Pro via the API. These are HD or better, not 480p. If your videos look lower quality, it may be due to app limitations, current rollout settings, or how the app encodes previews and downloads at this stage. There is no explicit mention in the sources that Sora 2 is “worse” than Sora 1 in terms of technical resolution—on the contrary, Sora 2 and Sora 2 Pro are described as more advanced and capable of higher fidelity, especially via API or Pro access Video generation with Sora Pricing.
-
App limitations: The Sora app (especially for Plus users) may restrict available resolutions or not expose all controls (like aspect ratio or number of outputs) that are available via the API or on the web. This could explain your experience of not being able to select resolution or aspect ratio Creating videos with Sora.
ChatGPT Pro vs. Plus Membership Differences
-
Pro users get higher quality and more features: ChatGPT Pro users can generate videos up to 1080p and 20 seconds, with up to 5 concurrent generations, and can download videos without a watermark (with some restrictions). Plus users are limited to 720p and 10 seconds, with up to 2 concurrent generations, and all downloads include a watermark Generating videos on Sora.
-
Watermarks: All Sora 2 videos have a watermark by default, but Pro users can download watermark-free videos if the video is generated from text, does not depict a public figure, and does not use someone else’s cameo. If you’re on Plus, all your downloads will have a watermark Creating videos with Sora.
Usability and Controls
-
Limited controls in the app: The Sora app currently does not allow users to select custom aspect ratios, resolutions, or the number of outputs per generation. These features may be available via the API or on the web, but not in the app interface at this time Creating videos with Sora.
-
Rollout and user experience: Sora 2 is still in a phased rollout, and some features (like higher resolutions, more control, and watermark-free downloads) are only available to Pro users or via the API. The app is designed for quick, collaborative creation, which may feel less “professional” compared to more advanced tools Getting started with the Sora app.
Summary:
Sora 2 is not technically lower resolution than Sora 1, but your experience may be limited by your ChatGPT plan (Plus vs. Pro), the app’s current feature set, and rollout phase. For higher resolution, more control, and watermark-free downloads, ChatGPT Pro or API access is required. The app is intentionally simplified for ease of use, which may not meet all professional needs at this stage.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
rare meteor
@outer basalt Why cant i make a moview with real human faces in sora 2 but i can in sora 1?
outer basaltBOT
Let me check the knowledge base for you again. One moment! :books:
You cannot make a video with real human faces in Sora 2 because OpenAI has implemented stricter guardrails and content restrictions in Sora 2 compared to Sora 1. Specifically, Sora 2 blocks the generation of videos that include real people—including public figures—or input images with human faces. This is a deliberate policy to prevent misuse, such as deepfakes or unauthorized use of likenesses, and to align with OpenAI’s updated safety and privacy standards Video generation with Sora Creating videos with Sora.
In Sora 2, the only way to use a real person’s likeness is through the “Cameo” feature, which requires explicit opt-in and verification from the person whose likeness is being used. Uploading images with depictions of real people is otherwise blocked Creating videos with Sora.
In contrast, Sora 1 did not have these same strict limitations, which is why you may have been able to generate videos with real human faces there. The change reflects OpenAI’s evolving approach to safety, privacy, and responsible AI use.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
rare meteor
@outer basalt isnt technology at a point where you can tell if i made up a character vs used a promonent person without limiting me to my own cameo and animals?
outer basaltBOT
Let me check the knowledge base for you again. One moment! :books:
OpenAI’s current policy for Sora 2 is to block the generation of videos with real human faces—including public figures—unless you use the Cameo feature, which requires explicit consent and verification. The knowledge sources do not provide technical details about whether the model can reliably distinguish between fictional characters and real people in user prompts or uploaded images. Instead, the policy is a broad restriction designed to prevent misuse, such as deepfakes or unauthorized use of likenesses, and to ensure privacy and safety Video generation with Sora Creating videos with Sora.
While technology is advancing, OpenAI’s approach is to err on the side of caution by limiting the use of real human likenesses to opt-in Cameos and non-human subjects, rather than relying on automated systems to distinguish between real and fictional people in every case. This is a policy choice, not necessarily a technical limitation, and is intended to align with OpenAI’s safety and privacy standards.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
rare meteor
@outer basalt then why not hve the same policy on all your products?
outer basaltBOT
Let me check the knowledge base for you again. One moment! :books:
OpenAI’s knowledge sources do not explicitly state why policies regarding the use of real human faces or likenesses may differ between products like Sora 1 and Sora 2, or why the same policy is not enforced across all products. The available information indicates that OpenAI regularly updates its policies to improve safety and adapt to new use cases, and that stricter guardrails have been implemented in newer products like Sora 2 to better protect privacy and prevent misuse, such as unauthorized deepfakes or impersonation. However, there is no detailed explanation in the sources about why legacy products may have different or less restrictive policies compared to newer releases Creating images and videos in line with our policies Usage policies.