> ## Documentation Index > Fetch the complete documentation index at: https://docs.wji.ai/llms.txt > Use this file to discover all available pages before exploring further. # Troubleshooting > What common errors mean and how to resolve them. ## Generation failures When a generation fails, WJI shows a plain-language reason rather than a provider error code. There are 33 of them, grouped below by what you should do about it. ### Content was blocked | Message | What to do | | -------------------------------------------- | --------------------------------------------------------------- | | **Contains a real person** | The reference image shows a real person. Replace it | | **Prompt is unsafe** | Rephrase in more neutral terms | | **Video / audio / image content restricted** | The *output* triggered a content limit. Adjust the prompt | | **Triggered safety policy** | Either the reference or the output. Change both if unsure | | **Contains prohibited content** | The image could not be produced. Adjust the prompt or reference | | **May involve copyright** | Change the prompt or the reference material | | **Output audio may involve copyright** | Adjust the prompt, dialogue, or reference audio | Content blocks are decided by the model provider, not by WJI. Rewording usually helps more than retrying. ### Reference material is wrong | Message | What to do | | ------------------------------------ | -------------------------------------------------------------------------------------- | | **Reference image too small** | The short side must be at least **300 px** | | **Audio too long** | Shorten it — Seedance generally needs 15 seconds or less | | **Audio too short** | Seedance 2.0 reference audio generally needs at least 1.8 seconds | | **Audio format unsupported** | Convert it | | **Video duration invalid** | Check the model's requirements | | **Edit video duration out of range** | Video editing needs a reference clip of **4 to 30 seconds** | | **Resolution too low** | Use a higher-resolution reference video | | **Missing reference material** | Add at least one reference image or video | | **Reference conflict** | Speech generation cannot take a reference image and reference audio together. Keep one | ### The prompt is the problem | Message | What to do | | --------------------------------------- | ------------------------------------------------------------------------------------------------------------------------ | | **Prompt missing** | Enter one | | **Prompt too short** | Add more description | | **Prompt does not match the task type** | A generate-mode prompt contains edit or extend keywords. Switch the task type, or reword | | **Missing edit intent** | Edit mode needs words like *edit*, *add*, *remove*, or *replace* | | **Missing extend intent** | Extend mode needs words like *extend* or *continue* | | **Task type mismatch** | Check task type, aspect ratio, duration, and prompt together — edit and extend need adaptive ratio and a reference video | ### Nothing came back | Message | What to do | | ---------------------------- | -------------------------------------------- | | **No image generated** | Adjust the prompt or reference and try again | | **No result from the model** | Same | | **Generation timed out** | Retry later | ### Service problems | Message | What to do | | ------------------------------------------- | ------------------------------------------------------- | | **Service error** | Temporary. Retry shortly | | **Quota exhausted** | Provider capacity, not your credits. Retry later | | **Network error** | Check your connection and retry | | **Invalid request** / **Invalid parameter** | Check the prompt and references | | **Generation failed** | The generic fallback when nothing more specific applies | ## Other common problems **"Cannot resolve asset ID, add the image to the asset library first"** — you dragged an image WJI could not identify. Save it to the asset library, then drag it from there. **A generation is running and I want to stop it** — you cannot. There is no cancel. Wait for it to finish or fail. **My references disappeared after switching models** — the new model accepts fewer, so the oldest were dropped. See [Choosing a model](/image-generation/models). **My result is stylized and I did not ask for it** — a style preset is still selected. It is appended to the final prompt without appearing in the prompt box. See [Writing prompts](/image-generation/prompts). **Undo will not bring back what I deleted** — check what undo covers. Creating, pasting, duplicating, uploading, and generation results are not undoable, and undo is cleared when you switch canvases. See [Autosave and undo](/infinite-canvas/autosave-and-undo). **I cannot upgrade my subscription** — there is no upgrade path. Cancel, wait for the period to end, then subscribe to the plan you want. See [Subscriptions](/billing/subscriptions). **"Workspace has reached its total / monthly limit"** — a spending cap is set. See [Workspace settings](/workspaces/settings). **"You are not a member of this organization" or "the organization is disabled"** — you are returned to your personal space automatically. See [Switching context](/organizations/switching-context). **I cannot remove a member from my organization** — they are still the budget owner of a workspace. Reassign it first. **The invite link I shared needs to stop working** — workspace links can be revoked; organization invitations cannot, and expire after 7 days.