AI Video Generation
Writingmate's Video workspace creates short videos from a text prompt and, when the selected model supports it, a reference image. Video generation is separate from chat and Compare mode.
Create a video from text
- Select Video in the sidebar.
- Describe the scene, subject, action, camera movement, lighting, and mood.
- Select a video model.
- Select Wide (
1280x720) or Tall (720x1280). - Select an available duration.
- If an audio button appears, use it to enable or disable audio for that model.
- Select Create.
Select Suggest if you want Writingmate to place an example idea in the prompt field. Review and edit the suggestion before creating a video.
Animate a reference image
- Select Image.
- Upload a JPEG, PNG, or WebP image.
- Describe the motion, camera behavior, and changes you want to see.
- Choose the model, output orientation, and one of the durations still available after the image is attached.
- Select Create.
Writingmate prepares the image for the selected output size, which can crop its edges. Check the preview and choose Remove if you want to return to text-to-video.
Reference-image support is model-specific. The models currently visible in the main selector are configured for image-to-video, but this can change when the catalog changes. Images with people, especially children, may be rejected by a model provider's content-moderation rules.
Models and conditional options
The model selector is built from Writingmate's current video catalog. The visible list can change without a documentation release. At the time of this update, it includes OpenAI Sora 2, Google VEO 3.1, Seedance 2.0, Kling 3.0, Kling 2.6, Kling O1, and PixVerse 5.5.
The form recalculates valid durations and resolutions when you change the model or attach an image. Current duration choices are:
| Model | Text-to-video durations | With a reference image |
|---|---|---|
| OpenAI Sora 2 | 4, 8, or 12 seconds | 4, 8, or 12 seconds |
| Google VEO 3.1 | 4 or 8 seconds | 8 seconds |
| Seedance 2.0 | 5, 8, or 10 seconds | 5, 8, or 10 seconds |
| Kling 3.0 | 5, 8, or 10 seconds | 5, 8, or 10 seconds |
| Kling 2.6 | 5 or 10 seconds | 5 or 10 seconds |
| Kling O1 | 5 or 10 seconds | 5 or 10 seconds |
| PixVerse 5.5 | 5, 8, or 10 seconds | 5, 8, or 10 seconds |
Audio behavior also depends on the model. VEO generates audio automatically and does not show an audio switch. Sora does not show an audio switch. Seedance, Kling 3.0, Kling 2.6, and PixVerse can apply the form's Audio enabled or Audio disabled choice. The current form can show that button for Kling O1, but its provider route does not apply the setting, so do not rely on controllable audio for Kling O1.
Some catalog entries describe capabilities such as end frames, motion control, additional aspect ratios, or multiple references. Those controls are not available in the main Video form today. Use only the options that appear after you select the model.
Track a generation
After Writingmate accepts the request, it shows Video generation started and adds a card to the Video Library. Jobs move through these states:
- queued: the request is waiting to start;
- in_progress: generation is running and may show a progress percentage;
- completed: the video is ready to play; or
- failed: the provider or processing step did not finish successfully.
The page checks active jobs in the background. Generation can take several minutes, and provider progress is an estimate. You can have at most three videos queued or in progress at once.
For a completed video, select its card to open the player. Use the card menu to Rename, Download, or Delete it. Delete cannot be undone. The Share button creates a public link, so review the video and prompt before sharing it.
When a video fails
Open the failed item in your video library and read the displayed error before retrying. Writingmate distinguishes provider-policy rejections from technical failures such as timeouts, networking problems, quota limits, and storage errors.
For example, FAL may reject a reference image when it appears to contain the likeness of a real person or other private information. Retrying the same image or changing only the prompt will not resolve an image-validation error.
See Video Generation Errors and Provider Policies for error-specific steps and a link to FAL's current Acceptable Use Policy.
Plan and AppSumo limits
Standard subscriptions use a per-video monthly allowance:
| Workspace plan | Monthly video allowance |
|---|---|
| Basic | 0 |
| Premium | 0 |
| Pro | 3 videos |
| Ultimate | 30 videos |
Queued, in-progress, and completed jobs count toward the standard monthly allowance. Failed jobs are excluded after their status is updated. Standard limits reset by calendar month.
AppSumo workspaces use a monthly pool of video seconds instead of a number of videos:
| AppSumo tier | Video seconds per month |
|---|---|
| Tier 1 | 30 seconds |
| Tier 2 | 120 seconds |
| Tier 3 | 300 seconds |
| Tier 4 | 600 seconds |
Each AppSumo request reserves the selected duration from the workspace pool. If the upstream generation fails, Writingmate returns those reserved seconds. AppSumo usage resets on the plan's monthly activation anniversary. See AppSumo Plan Tiers and Compare Subscription Plans.
Write a useful prompt
Include the details that must remain consistent:
- subject and setting;
- action and direction of movement;
- camera framing and movement;
- lighting, color, and mood; and
- audio, dialogue, or ambient sound when the selected model supports it.
For image-to-video, describe the motion rather than repeating what is already visible in the image. Generated output can contain visual errors, unexpected motion, or inaccurate text. Review it before publishing or using it in customer-facing work.
Prompts, reference images, and generated videos are processed by Writingmate and the selected model provider. Do not upload confidential material or media you do not have permission to use.