- Browse the Discover section first — Go to the Discover Content page (Earth icon on mobile). See other users' full generation settings — prompts, extensions, models — then replicate with one click and modify from there. This is the fastest way to learn the platform.
- Choose the right prompt style — Use Simplified mode mainly for base models and base-extension workflows. RAW Photo 2 and Photoreal models usually work better with natural language, while anime and hentai models usually work better with short tags.
- Use natural language for videos — For Animate V2.1 and V4, write a natural-language action prompt. Use the selected extension's Recommended Keywords / Tags as wording cues, especially for sex-scene extensions.
- Use Templates — The Templates button on the generator homepage provides prompts pre-tuned for each AI model. Especially useful when you are learning a new model family.
- Match the model and extension family — Base, older HD, HD 2.0, RAW Photo 2, Velora, and Animate models use different extension families. If an extension does not appear, the selected model is probably not compatible with it.
- Keep Other short in Simplified mode — In Simplified mode, the Other section works like a small prompt injector. Keep it focused on a few extra details; stuffing it with too many words can dilute the selected extensions.
- Do not select conflicting extensions — Two mutually exclusive poses simultaneously will cause malformations. If generations are looking broken, conflicting extensions are the most likely cause.
- Use upscaling correctly — For base and older HD generations you want to keep, try 2× upscaling before saving or sharing. For HD 2.0 / RAW Photo 2, 2× and 3× use a different upscale path, so choose based on the final resolution you need.
Image Generation
Tips & Best Practices
Work in progressWe are still adding and refining Help Center articles.
Still need help?
Use the support center for account, billing, or platform help, or email us directly.