Participating Frequently

Answered

refuses to animate an image of a hand drawing in the sand

Forum|Forum|17 days ago
July 13, 2026
5 replies
46 views

The image is a man crouched down and drawing in the sand with his finger. The face is not seen. The prompt is “the man draws in the sand with his finger.” I’ve tried variations of this prompt like “the hand draws in the sand” or “the man doodles in the sand” but Firefly continues to refuse to animate the video. It’s not giving a community guidelines error (and it shouldn’t since nothing about this image or prompt violates guidelines) but it’s flatly refusing to generate the video.

Correct answer Hajar Toutsi

Thanks for all the detail you sent over Doc_K, the screenshots and the reference image made it much easier to pin down what's actually happening here.

This one turned out to be on the model's side rather than ours. Veo processes each prompt through its own validation before generation even starts, and Firefly doesn't have visibility into why a specific prompt gets rejected there. That's why the same setup works fine on a different subject and fails here, even with the seed and reference image changed.

I know that's not a satisfying answer when you've already ruled out every variable on your end. Unfortunately we can't see the reasoning behind Veo's rejection, so I can't tell you what specifically to adjust to get this one through.

If you're open to it, trying a different partner model for this particular clip might get you unstuck faster than continuing to iterate on the same prompt.

Hajar

Trevor.Dennis

Community Expert

BTW I asked it to have the man use his finger, and got him using this stick or rock or whatever it is.

Trevor.Dennis

Community Expert

I’d be using a more wordy prompt with some redundancy. If there is the slightest chance that the model will find any part of the prompt confusing and ambiguous, then take it for granted that the model will do exactly that. Try:

A man is crouching on a sandy beach. He uses his right hand with an extended index finger, and drags it through the sand to create the following text: ‘your text here’.

If you have Photoshop or similar then consider creating the last frame with the text. You could use Gemini 3.1 to create the first frame, and your skills to create the last.

In case you don’t know, the way to create indented text/anything is to use Bevel & Emboss with Direction set to Down and Fill Opacity set to zero.

Is your end product video or frame animation?

Hajar Toutsi

Community Manager

Hi Doc_K,
Thanks for reaching out about this.

If you are encountering a specific error message or the interface is not behaving as expected, please provide the following:

A Full Screenshot: Please provide a screenshot of your entire browser or application window, including your avatar in the upper right corner. Many errors have multiple causes; seeing the full context is much more helpful than a zoomed-in image of the error message alone.
System Details: Specify your Operating System (OS) and Browser (including version numbers).
Workflow Context:
- Which model was selected?
- Copy of the reference image.

Once we have the details, we’ll be glad to help guide you get the results you aim for.

Hajar
[Moved to questions]

Doc_KAuthor

Participating Frequently

Apologies for the delay. I’m very busy, and delays like this make things worse. I’ve attached a full screenshot from the attempt from this morning on an entirely different machine because it’s still giving the same response.

I’ve added a screenshot of the complete info on the system I’m currently on, the same one the Firefly screenshot was taken on.

I’ve also attached the exact image I’m trying to animate, the one the model says it can’t load, so you can see the full image.

Immediately after this failure, I was able to generate a video of an entirely different image (a woman in a rocking chair), so it’s not that the model isn’t working.

To be clear, I use the exact same model every time I generate, and I don’t usually have an issue unless a child is in the image.

Hajar Toutsi

Correct answer

Community Manager

Thanks for all the detail you sent over Doc_K, the screenshots and the reference image made it much easier to pin down what's actually happening here.

This one turned out to be on the model's side rather than ours. Veo processes each prompt through its own validation before generation even starts, and Firefly doesn't have visibility into why a specific prompt gets rejected there. That's why the same setup works fine on a different subject and fails here, even with the seed and reference image changed.

I know that's not a satisfying answer when you've already ruled out every variable on your end. Unfortunately we can't see the reasoning behind Veo's rejection, so I can't tell you what specifically to adjust to get this one through.

If you're open to it, trying a different partner model for this particular clip might get you unstuck faster than continuing to iterate on the same prompt.

Hajar

Sign up

To post, reply, or follow discussions, please sign in with your Adobe ID.

Sign in to Adobe Community

To post, reply, or follow discussions, please sign in with your Adobe ID.