Troubleshooting

Fixes for the banners and errors you're most likely to see.

This Runpod machine’s GPU isn’t working

A red banner with this message means the GPU on the machine Runpod gave you can’t be used. It happens now and then on Community Cloud machines, and nothing inside the pod can fix it.

Fix: in the Runpod console, terminate the pod and deploy the template again. You’ll land on a different machine. If you already made work on this pod, back it up first. Deploying on Secure Cloud almost always avoids this.

This pod has no storage volume

The pod was deployed without a volume, so the models won’t fit and your work would be lost when the pod stops. The studio waits instead of starting the download.

Fix: terminate the pod and deploy again. On Runpod’s deploy page, click Add volume (180 GB) before Deploy Pod.

Your volume is too small for the models

The volume doesn’t have room for the models that are still downloading. The banner shows how much is free and how much more is needed.

Fix: stop the pod, then in Runpod use Edit Pod to make the volume bigger, and start it again. The download picks up where it left off.

Downloading models

The amber banner at the top shows models still downloading after the first boot. Features unlock as their models arrive. Until then their Generate button is disabled, and hovering over it shows Model files still downloading.

Everything is usually ready in about 10 minutes. Check progress any time under Settings → Models.

A model download is waiting for a Hugging Face token

Some models, including LTX-2.5 (the default video engine), are gated: you have to accept their terms on Hugging Face before you can download them.

Fix: the banner says “Video needs a free Hugging Face token.” — click Set up video. The dialog walks you through it: accept the terms, create a read token, and paste it in. The download starts within about 15 seconds. That field is also under Settings → Models, if you’d rather use that.

Still waiting after you saved a token? The banner then says what Hugging Face objected to, and Fix it shows the file, the repo and your account:

• Rejected the saved token: it’s mistyped, revoked or expired. Create a new Read token and save it.
• Refused a file for your token: that account hasn’t accepted the terms for this exact repo (accepting an older LTX repo doesn’t count), or it’s a fine-grained token without Read access to contents of all public gated repos you can access. Fix that on Hugging Face, then click Retry now.

A mode or page is missing

The studio hides features your pod has no models for. This only happens if some model downloads were switched off when you deployed.

Fix: check Settings → Models. Groups marked Not enabled were switched off. Turn one on by setting its DOWNLOAD_*_MODELS variable to true with Edit Pod in Runpod.

Setup is locked

A new studio can only be claimed in the first 15 minutes after the pod starts. If nobody created a password in that time, setup locks.

Fix: do one of these:

• Restart the pod from the Runpod console. Setup reopens for another 15 minutes.
• Click Edit Pod in Runpod and set STUDIO_PASSWORD to the password you want, ideally as a Runpod Secret.

Your models and films are kept either way.

I forgot my password

In the Runpod console, click Edit Pod and set STUDIO_PASSWORD to a new password. Saving restarts the pod, and the new password replaces the old one. Use a Runpod Secret so the password isn’t shown in plain text in the console.

After 10 sign-in attempts in a minute, the studio stops checking passwords for up to a minute, and the login screen keeps saying Incorrect password even when the password is right. Wait a minute before trying again.

Break down with AI is missing

Script breakdown needs an AI key. Set ANTHROPIC_API_KEY (or an OPENAI_* provider) as a pod variable with Edit Pod in Runpod — see Settings and API keys. Until you do, you can still build a film by hand.

A job failed

Open the Queue at the top right. Failed jobs show their error, and you can click Retry.

Storage is almost full

When less than 3 GB of storage is free, the studio refuses new jobs with The pod’s storage is almost full.

Fix: back up what you want to keep, then delete old images and videos from the gallery. If you keep running out, deploy a new pod with a larger volume.

A Civitai LoRA won’t import

Some Civitai files, including all NSFW ones, need a Civitai token. Add one under Settings → Download tokens.

Still stuck?

Ask in the Blockbuster Studio Discord, or open an issue on GitHub with what you were doing and the exact message you saw. Settings → Setup and logs → Download logs gives you a file to attach; it has no passwords or keys.