RIP Gemini Image Generation
As someone who used Gemini for product images and clothing specifically, a bit over a month ago it lost the ability to create realistic images without morphing labels. (Plus a plethora of other issues)
Here’s what I’ve learned since:
- I used to drop in a ton of reference photos to accurately show a certain aesthetic or vibe, now it will copy one of the images. It struggles specifically with angles and camera placement.
- ChatGPT has gotten better at images for my specific use, however I feel that the image resolution isn’t the best and once you edit it, all images will have weird marbled lines especially with hair.
- Google Flow Labs sometimes works accurately, however it has nonsensical guardrails and will hallucinate
Anyone found a workaround to get the old Gemini’s skills back? Or can recommend a substitute?