LocalGen: AI Art Playground

Powerful and quick model. Definitely better than apple's image playground junk. I bought the pro license! I love the app. However, I dislike that after every iOS update that each and every model installed has to be re-compiled at a cost of ~200sec each (10min+). For me, that means each time I use the app after iOS update I do not get to use the app while the idea is fresh in my mind. I forget what I was gonna to create. I then spend the time re-compiling. This is a frustrating process and it takes me a while to want to use the app again. I have PTSD and it hurts to continually wait 10 minutes re-compiling. It is so frustrating that you have to do all this when you want to use an app. I appreciate the developers response and I have updated my review accordingly.

Hello developers! First, let me say that I really enjoy how straight forward this app is, in fact it’s ease of use is what made me decide to make a HUGE $30 purchase even though the Draw Things app can do text to image for free and offers all text-to-image models on their platform (just search & download). My main issue is that I downloaded all 5 models offered in this app, then I came back to it later and it wanted me to re-download the models again. This is a lengthy process and I really just wanted the models to be stored so I could switch between them. Why did the models get flushed out and require me to re-download them?

There are not enough apps like this on the store for local generation, and this one is mostly free! You can generate unlimited images quickly which is great. I like there are no ads or subscriptions, just a reasonable 1-time purchase if you want a bit more functionality. It could use some enhancements. The image quality is too low to be used for anything other than novelty. Additionally, the model does not react to detailed prompts. It might get the right subject, but struggles with details like “subject in x setting”. Getting a person or animal to reliably pose a certain way or act a certain way is difficult and often results in extra limbs, disfigurement or just general weirdness, jf it doesn’t ignore the details altogether. So additional improvements to how it handles detailed prompts is needed. I’d wait longer for a beter output that more accurately matches my prompt. It also cannot handle text. If the dev can get Flux working instead of SDXL, I think that may also lead to better results… but I know that may be difficult. Overall, excited to see how updates improve this app!