# Reference Images for Veo-3

**URL:** <https://community.mindstudio.ai/t/reference-images-for-veo-3/1203>\
**Category:** Feature Requests\
**Created:** [July 6, 2025, 12:03am UTC](https://community.mindstudio.ai/t/reference-images-for-veo-3/1203 "2025-07-06T00:03:35Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Ken](https://yyz2.discourse-cdn.com/flex008/user_avatar/community.mindstudio.ai/ken/32/493_2.png) [@Ken](https://community.mindstudio.ai/u/Ken)\
**Post date:** [July 6, 2025, 12:03am UTC](https://community.mindstudio.ai/t/reference-images-for-veo-3/1203/1 "2025-07-06T00:03:35Z")

</div>

I’m delighted that you’ve recently added Veo-3 to the “Generate Video” block! Veo-3 supports the use of reference images, but there doesn’t appear to be a way to specify these in the model configuration section. Can that be added, for up to two images? Apparently, Veo-3 requires image file uploads, and doesn’t support URL image references. The Veo-3 API supports direct image upload as a **base64-encoded byte string**.

There are a lot of useful uses purposes for references images; for example:

- Visual style reference
- Character consistency across multiple video clips
- Animatable inspiration
- Frames-to-video (start and end images for an animation or transition sequence)
- And more…

Thanks!

---

<div class="post-metadata">

**Author:** ![ryanwhitaker](https://yyz2.discourse-cdn.com/flex008/user_avatar/community.mindstudio.ai/ryanwhitaker/32/1439_2.png) [@ryanwhitaker](https://community.mindstudio.ai/u/ryanwhitaker)\
**Post date:** [October 15, 2026, 3:20am UTC](https://community.mindstudio.ai/t/reference-images-for-veo-3/1203/2 "2026-10-15T03:20:36Z")

</div>

+1 on this. Until reference images are exposed in the Generate Video block, one workaround is to call the Veo API from a custom function: fetch the image from its URL inside the function, convert it to base64, and pass it as the image input in the request. That gets around the URL-vs-upload limitation you mentioned.

Character consistency across clips is probably the use case that benefits most, since prompt-only descriptions tend to drift between generations. Optional start and end frame inputs on the block would also cover a lot of transition sequences.

---

<div class="post-metadata">

**Author:** ![Alex\_MindStudio](https://yyz2.discourse-cdn.com/flex008/user_avatar/community.mindstudio.ai/alex_mindstudio/32/27_2.png) [@Alex\_MindStudio](https://community.mindstudio.ai/u/Alex_MindStudio)\
**Post date:** [October 15, 2026, 1:01pm UTC](https://community.mindstudio.ai/t/reference-images-for-veo-3/1203/3 "2026-10-15T13:01:56Z")

</div>

Hi @ryanwhitaker, welcome to the community!

Reference images, as well as start and end frames, are now available with the Veo 3.1 models in the Generate Video block
