PixelDojo updates photo effects and restores text-only video

A photo effect has a simple promise: the person you start with should still look like that person once the picture moves. PixelDojo’s October 30 update changes the engine behind its video effects, with claimed improvements to the opening frame and face consistency, including two-person effects. Dated release note.

The same update says MiniMax H3 Turbo can generate from text alone again; a starting photo is optional. These are two different starting points. Use a photo when the existing subject matters. A text-only request leaves more of the subject and composition for the model to invent.

Check what the motion changes

PixelDojo describes a closer match to the supplied photo at the start and better preservation of both people in two-person effects. Those are vendor claims, not results from a generation test we ran.

For a proposed review, compare the opening, middle and final frames with your original. Look for a changed face, clothing detail or hand position that the motion does not explain. In a two-person shot, inspect both subjects rather than judging only the more prominent one. A convincing first frame does not establish a consistent whole clip.

If you are exploring an imagined scene instead, write down the action and camera movement before trying text-only generation. Keep one version you can compare against; changing the subject, action and camera together makes it harder to identify what helped.

Choose the right allowance before generating

The current pricing page lists three free video or photo effects per day with a small watermark. Paid plans start at $10 monthly and include selected unlimited Relax tools, including MiniMax H3 Turbo and video effects. Other models continue to use credits; this is not unlimited access to every generator.

The effects page describes the photo-based starting point. The dedicated MiniMax page redirects to sign-in, so account-specific access, charges outside the listed Relax mode and rollout by region were not independently verified.

Begin with one effect or one short text-led idea, inspect the whole clip and confirm the mode shown in your account before running variations. Our AI output evaluation guide helps make those checks specific. The earlier film-planning story covers assembling a sequence, which still needs a separate continuity review.