AIFFast

Copy your recovery link or download your session file before you spend credits.
Pick the correct mode first: Generate, Edit, Video, or Image video.
Set the right aspect ratio before you queue the job. This prevents avoidable reframing.
Write or generate a prompt, review the credit cost, then submit.
Use the Results area to download, continue, or stitch finished clips.

Your session token is your access key. If you lose it, you lose access to that session’s credits and results.
Use Copy recovery link for fast restore on the same or another device.
Use Download session file for a local backup you can import later with Restore file.
Files stay available for 6 days after completion, then they are removed automatically.
Top ups add credits to the current session only, so make sure you are in the correct session before paying.

Fast costs 1 credit and is the low-cost general lane.
Explicit costs 2 credits and is the direct adult image lane.
Style only applies to Fast. Fast supports photoreal, anime, and 3D cartoon. Explicit text-to-image remains photoreal only.
Size changes aspect only. Pick the frame shape you actually want before generating.
Best prompt shape: subject, action, scene, mood, and any important framing details in plain language.
Do not cram many unrelated ideas into one image prompt. One clear scene usually performs better than a laundry list.

Source image is required. Use a clear image that already matches the composition you want to preserve.
This mode currently costs 3 credits per generation.
Source image look tells the app what the uploaded image currently is.
Output look tells the app what to convert the source into.
This route is best for look conversion and scene adjustment, not for inventing an unrelated replacement image.
The prompt is optional in this mode. Use it only to add extra steering on top of the source image and selected conversion route.
For stronger preservation, keep the source image close to the target aspect and avoid vague prompts that imply a brand new scene.

Normal and NSFW both use the fixed video path for the chosen aspect.
Text-to-video currently costs 6 credits per generation.
There is no style selector here. This video lane is routed as photoreal video.
Start frame is optional. Leave it empty for pure text-to-video, or upload one when you want the clip to begin from a specific frame.
Video ending controls whether the clip stays open-ended or is driven toward a supplied last frame.
No end keeps the clip open and does not lock an end frame. This is the right choice when you want a continuation-friendly segment.
Start + end requires both a start frame and an end frame, and the model tries to travel between them while following your prompt.
Write prompts as short motion descriptions: who or what is in frame, what is happening, where it happens, and how the camera should feel.
Best results usually come from one scene, one action, and one visual direction instead of a long multi-scene paragraph.
If the result is close but needs a next beat, use Continue video rather than rewriting the whole clip from scratch.

Start frame is required. Use a clean frame with the subject already centered the way you want the clip to begin.
This mode currently costs 6 credits per generation.
Size selects the animation path by aspect, so choose the option that matches the uploaded image best.
Video ending controls whether the clip stays open-ended or is driven toward a supplied last frame.
No end keeps the clip open and does not lock an end frame. This is the right choice when you want a continuation-friendly segment.
Start + end requires both a start frame and an end frame, and the model tries to travel between them while following your prompt.
Guidance controls how strictly the motion follows the text prompt. Lower values stay closer to the source image. Higher values obey the prompt more aggressively.
Use this mode for animating an existing frame, not for changing the core identity, outfit, location, or camera layout unless that change is intentional.

Promptgen costs 1 credit and writes a fresh prompt for the current mode.
NSFW Promptgen costs 1 credit and writes a prompt for the explicit lane. Use it when you want a generated explicit prompt instead of writing one yourself.
Refine prompt costs 1 credit and rewrites your current prompt for the selected mode.
Generate first when you need a new idea. Refine when your existing idea is right but the wording needs to be cleaner.
Always review prompt-tool output before submitting. It is a drafting tool, not an automatic guarantee.

Download Original saves the raw file exactly as generated.
Finished video cards also expose Download MP4 and Download GIF.
Continue video costs 8 credits and starts the next clip from the last frame of the selected video.
Edit prompt + continue lets you branch the next segment with a rewritten continuation prompt before queueing it.
Select for stitch marks finished clips for combination. Stitching costs 1 credit per extra clip after the first one.
For the cleanest stitched result, use clips with the same aspect ratio and a natural sequence order.

Pick the mode first. A good prompt in the wrong mode still gives the wrong kind of result.
Match your aspect ratio to the shot you want. Portrait for tall subjects, landscape for wide scenes, square for centered compositions.
Use prompts that describe one clear result, not several competing results.
For edits and image-video, use the cleanest possible source image. Bad framing in means bad framing out.
Use continuation when a clip needs one more beat. Use stitching when you already have several finished clips that belong together.

Generate images, edit uploads, and render text or image driven video in one session.