#Ability to hold/lock character models

1 messages · Page 1 of 1 (latest)

gilded imp
#

What’s OpenAI’s agenda when it comes to sora 2, with locking character models.
So we can generate content that is consistent?

quiet sandalBOT
#

Hi @gilded imp. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

#

I'm sorry, but the knowledge sources do not contain explicit information about OpenAI providing a feature in Sora 2 for "locking" or "holding" character models to ensure consistent character generation across multiple videos. The documentation does mention that you can guide video generation with an input image (used as the first frame), which can help preserve the look of a brand asset, character, or environment for a single video, but it does not describe a system for persistent or reusable character models across projects or sessions. Additionally, there are restrictions on generating videos with real people or faces in input images, which may further limit character consistency options at this time Video generation with Sora – Use image references.

If you are interested in cameo features (where you or approved friends can appear as characters), these are opt-in and permissioned, but again, this is not described as a general-purpose character locking system for arbitrary fictional characters Getting started with the Sora app.

In summary, based on the available documentation, OpenAI has not announced or documented a feature for locking character models for consistent content generation in Sora 2.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@gilded imp Kapa.ai is still learning and improving, please let me know how I did by reacting below.

gilded imp
#

@quiet sandal so that’s a no to the future of sora having such features?

quiet sandalBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

Based on the knowledge sources provided, there is no explicit information about OpenAI’s future plans to add a feature in Sora 2 (or future versions) that would allow users to lock or hold character models for consistent content generation across multiple videos. The documentation only describes current features, such as using input images for single-video guidance and cameo functionality for representing real people with permission, but does not mention persistent or reusable character models as a roadmap item or planned feature Video generation with Sora – Use image references Getting started with the Sora app.

Therefore, I cannot confirm whether such a feature will or will not be available in the future, as the knowledge sources do not address this topic.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@gilded imp Kapa.ai is still learning and improving, please let me know how I did by reacting below.

gilded imp
#

@quiet sandal can you give me the road map for the company

quiet sandalBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

I'm sorry, but the knowledge sources do not contain a detailed or official roadmap for OpenAI as a company. The available information primarily covers recent and upcoming product releases, feature updates, and API changes, but does not outline a comprehensive long-term roadmap or strategic agenda.

If you are interested in recent and upcoming product directions, OpenAI has recently focused on:

  • Advancing reasoning models (o-series, GPT-4.1, GPT-5, etc.)
  • Expanding multimodal capabilities (text, image, audio, and video generation, e.g., Sora 2 and Sora 2 Pro)
  • Improving developer tools and APIs (e.g., Responses API, Realtime API, Agent Builder, ChatKit)
  • Enhancing fine-tuning, preference optimization, and evaluation tools
  • Increasing support for enterprise and education use cases
  • Adding features for content safety, watermarking, and compliance

However, for a formal or forward-looking company roadmap, the knowledge sources do not provide enough information to answer your question definitively.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@gilded imp Kapa.ai is still learning and improving, please let me know how I did by reacting below.

distant pier
#

@quiet sandal
how can i set base64 encoded png as input reference for sora 2

quiet sandalBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

Based on the knowledge sources, to set an input reference image for Sora 2, you must provide the image file as the input_reference parameter in your POST /videos request. The documentation specifically shows using a file upload (e.g., with curl’s -F and @file syntax) and lists supported formats as image/jpeg, image/png, and image/webp. However, it does not mention support for providing a base64-encoded image directly as the input_reference.

All available examples use file uploads, not base64 data URLs. For example:

curl -X POST "<https://api.openai.com/v1/videos"> \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: multipart/form-data" \
  -F prompt="..." \
  -F model="sora-2-pro" \
  -F size="1280x720" \
  -F seconds="8" \
  -F input_reference="@sample_720p.jpeg;type=image/jpeg"

There is no documentation or example showing how to set a base64-encoded PNG as the input reference for Sora 2. Therefore, based on the current knowledge sources, you must upload the image file directly, and providing a base64-encoded image as input_reference is not supported at this time Video generation with Sora – Use image references.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@distant pier Kapa.ai is still learning and improving, please let me know how I did by reacting below.

distant pier
#

how can i use image file as input reference for sora 2 with nodejs