# lora.pro

> AI photo and video generation from your own photo. Upload a face photo once,
> pick a style from the catalogue (or write your own prompt), and get a styled
> shot in seconds — or animate that shot into a short video with sound.
> Works on the web and as a Telegram Mini App. Interface in Russian and English.

## What the platform does

- **Photo generation from your photo.** The face from your uploaded photo is
  preserved while the scene, outfit, lighting and styling come from the chosen
  prompt. Several photos can be uploaded to build a character profile.
- **Style catalogue.** Thousands of ready prompts grouped by subject, mood and
  use case, with solo, duo and group variants. Every prompt page shows example
  results and the model it was tuned for.
- **Custom prompts.** Any text prompt can be run against the same pipeline.
- **Video generation from a photo.** A still frame (an uploaded photo, a saved
  character, or a previous generation) is animated into a short clip with
  sound. Duration, aspect ratio and quality are picked per model.
- **Multi-character video.** Some models accept several reference photos, so
  more than one person can appear in the same clip.
- **Editing and reuse.** A finished generation can be refined, repeated with
  other settings, or published to the public gallery under an author profile.
- **Tokens.** Generation is paid in tokens; new accounts get starter tokens for
  free. Prices per model are listed below and on https://lora.pro/prices.

## Photo models

- **Seedream 4.5** — 5 tokens per image; ~37s average
- **Banana Pro** — 9 tokens per image; ~20s average
- **Banana Flash** — 5 tokens per image; ~15s average
- **GPT-2 Лучшее качество** — 26 tokens per image; ~100s average
- **GPT Image 1.5** — 7 tokens per image; ~63s average
- **GPT-2 Среднее качество** — 2 tokens per image; ~36s average

Token prices are per generated image and already include the model's own cost.

## Video models

- **Kling 3** — 5–15 s; hd 23 tokens/s; multi-reference
- **Grok 1.5** — 4–15 s; standard 19 tokens/s, max 34 tokens/s
- **Veo 3.1** — 4–8 s; 1080p 20 tokens/s, 4k 48 tokens/s; aspect 9:16 / 16:9
- **Grok 1.5 Мульти** — 8 s; 480p 11 tokens/s, 720p 19 tokens/s; aspect 9:16 / 16:9 / 1:1; multi-reference
- **Seedance 2.5** — 4–30 s; 480p 30 tokens/s, 720p 64 tokens/s, 1080p 145 tokens/s; aspect 9:16 / 16:9 / 1:1; multi-reference
- **Seedance 2.0** — 4–12 s; 720p 41 tokens/s, 1080p 93 tokens/s; aspect 9:16 / 16:9 / 1:1; multi-reference
- **Seedance 1.5** — 4–12 s; 720p 7 tokens/s, 1080p 16 tokens/s; aspect 9:16 / 16:9 / 1:1

Video price = tokens per second × chosen duration. Aspect ratio never changes
the price; duration and quality do.

## Key pages

- [Home](https://lora.pro/): catalogue entry point, both generation modes
- [Create photo](https://lora.pro/create): photo generation
- [Create video](https://lora.pro/create/video): video generation
- [Prompt catalogue](https://lora.pro/prompts): browsable styles with filters
- [Pricing](https://lora.pro/prices): token packages and per-model cost
- [About](https://lora.pro/about): how the platform works
- [Experts](https://lora.pro/experts): prompt authors
- [Terms](https://lora.pro/terms), [Privacy](https://lora.pro/privacy), [DMCA](https://lora.pro/dmca)

## Notes for agents

- Russian and English are both first-class: `https://lora.pro/…` is English,
  `https://lora.pro/ru/…` is Russian.
- Model names above are the public names shown in the interface. The
  underlying infrastructure and routing are deliberately not published.
- Generation requires an account and tokens; there is no anonymous public
  generation API.
- Prices and the model list change — re-read this file rather than caching it
  for long periods.