Qwen Image 2.0
text→fotka
Poměry stran: 1:1, 9:16, 16:9
Cena: 68 kreditů
Silný v
- 20 billion parameters with a double read of the input — it grasps the meaning separately from holding the look of the image
- Commercial-grade text inside the image, including multi-line and whole paragraphs
- When editing, it KEEPS the font, the size and the fit of the existing text while it changes it
- First on all nine public benchmarks in its class
- An open model — predictable, with no surprises in the rules
Slabý v
- Only English and simplified Chinese are guaranteed — other languages, yours included, usually work, but it isn’t promised
- More restrained in style than the artistic models
- Slower than the fast ones
- Photorealism on people falls behind the leaders
Nepoužívej ho na
- Don’t count on it for text in your language — it isn’t guaranteed
- Don’t use it for a close portrait
- Don’t use it for a quick batch of test shots
Typické úlohy
- Editing text in an existing image while keeping the font
- An image with a multi-line caption in English
- A design with structured text
- A precise change to one element in the shot
