PirateDiffusion's new integrated WebUI Stable2go V2 is here

You can reply to a picture from PirateDiffusion telegram with /unified to launch the new web ui. To access your dashboard, send yourself a magic login link here: my.graydient.ai/switch. To learn all of the "other" commands you can send WebUI, check out piratediffusion.com/features

[@righthxvh6111]

Seeing the other reviews of this service so far the methods of generations are indeed "convoluted". Unlike other AI generators it's capability to generate HQ and accurate images is almost impossible.
First, the best part of this tool is most likely the sheer amount of 'concepts' but everything else is lacking with generations not sticking to prompts, lacking detail and more importantly builds coming out shoddy. Unlike other AI tools which require one image is fine for training the AI. This one needs multiple, HQ and accurate images but I still got nowhere near the generation I was hoping for. When using this generation websites I like to one use HQ images I've generated before to try keep same characters as well as generating new characters in which both these features using Stable2Go and Lora is IMPOSSIBLE. Even 'professional' concepts come out worse than expected. I kept perceiving but after 9 models built and several hours wasted, I gave up.
If you are looking for an AI generator, tester or multi-models avoid Graydient altogether. While they are a nice business on the surface, you can't sacrifice the service for kindness.

[@piratediffusion]

Thanks for the feedback, but how could it lack detail if its the same Stable Diffusion as anywhere else? The only possible difference is forgetting to turn up the resolution and using the right negative prompt. Have you tried this?
piratediffusion.com/free-image-upscaler-stable-diffusion-guide/
Also, you can ask any model trainer: 1 image "training" doesn't exist. Other services are disguising that as training but it is in fact IP Adapters V2. You can also do that with /blend.
Save a controlnet preset image of a face and then use the /blend command for 1-shot faces or use the faceswap/facepush method

[@righthxvh6111]

@@piratediffusion Hi.. thanks for the response, shows you value your user base. In general, with tweaking settings I found the results still came under similar software for both anime creation and realistic generations. As for the one shot face, I was indeed talking about a similar feature to your 'blend' feature, however unlike it, it can generate accurate images of the face inputted.
I don't know why this is an issue with this specific AI generator but AI characters are becoming universally more common even within sites like e-commerce clothes shopping sites (due to the face swap feature). Unfortunately my free trial I was using to experiment is indeed nearly over but if I have time tomorrow I'll test a bit further. Np. 👍🏼