surprised
I am usually very skeptical about image generation when it goes about getting precisely the results you need. Quality of images pretty good.
I don't like when the hairstyle is changed much without being requested. This wastes credits. But after a while I've learned to always mention the hairstyle and glasses to keep it consistent.
I used my 2000 credits within half day on experiments. And I went from " maybe I will refund" to " well maybe this is worse keeping".
Keep in mind that model generation costs 1500 credits. And each image is approximately 5. Many of them got wasted for me but still I managed to have almost 20 quite decent images. I think that for personal use 2000 credits per month is quite enough.
I didn't see any advanced image editing or retougching tools as declared in description. You create a model and then you write prompts. If you don't like the image or it is not perfect enough, you cannot edit it or improve it somehow. I just deleted it and modified the prompt to generate new ones. It is also not working to add a reference image and ask to add a minor change in it. It doesn't preserve the reference image. If I put " reference image style strengths" to maximum, it will not make the change asked (I asked to add glasses) but will change the image itself resulting in a little bit other result than expected and loosing details that I like in reference image.
Despite all these shortcomings I will give the product 5 tacos. Just because so far these are the best ai-generation based on selfie I tried
Creative_Bot_PhotoGPTAI
Aug 4, 2025Thank you so much for this generous review and for sharing your honest experience, we're genuinely grateful for your trust and feedback.
We completely understand the need for more refined control when it comes to retouching and targeted edits. Just to share a couple of tips that might help:
- If you're happy with a particular generation but want to tweak just one part, you can click the “Generate Similar” button. This will replicate the original image exactly using the same model, preset, and settings. You can then modify just the prompt (say, add or remove glasses, adjust hairstyle, etc.), and the system will retain most of the composition while reflecting your small change.
- That said, you're absolutely right that true retouching is still missing, and we're actively working on it. Our team is currently building an inpainting feature, which will let you select parts of the image and edit just those without disturbing the rest. This should make refining your images much easier.
We also have several other improvements planned in this direction. To get sneak peeks, share feedback, or get faster help, we’d love to have you join our Discord community: https://discord.gg/rENRJw439H.
As for the reference image feature, it's designed to help guide the style or composition, not to copy it 1:1. If you ever want help using it more effectively, feel free to drop us a line at support@photogptai.com or reach out in Discord.
We’re so glad you chose to stick with PhotoGPT and we’re here to help you make the most of it. Feedback like yours fuels how we grow, and it truly means a lot to the team.
Thanks again for the kind words and thoughtful suggestions!