AiPlaygroundWM

Up scale comparison test  original image Every scale size is the same.(two times) 1. using  https://www.iloveimg.com/ 2. using Local Stable diffusion WebUI  i2i ( Tiled Diffusion extensions)  4x-UltraSharp upscaler  denoising 0.4 3. using Local Stable diffusion WebUI  Extras tab , 4x-UltraSharp Using iloveimg site is free and no need local computing power. But the image quality is the best in using local Stable diffusion WebUI i2i 

While working on other tasks, I suddenly want to testing SDXL. I temporarily ran it once to make it easier to load the checkpoint and settings. I ran it with the basic setting of batch count 4, but it was still slow However, I realized that it was only 3 minutes per image. Thinking that it was doable, I forgot to reboot and ran it again It seems doable. At this point, I suddenly curious.  If I use 512 * 512 with hires 2 times instead of 1024 * 1024, will there be a difference? 512 * 512 size and hires 2 times, It takes 2min 18sec 1024 * 1024size once more 5min 7sec...?? It takes more time? When I turned on the task manager, I noticed that the SSD usage rate was rising to 100% along with the RAM.Hm....Interesting. Try again? (error) Fxxk!!!!!!!

Today, I will try the medvram argument test My computer is a laptop (MSI GP76) with a 3070 TGP140w and 8GB of VRAM So I cannot just use SDXL. (The model itself cannot be loaded.) The test model is  protovisionXLHighFidelity3D (It is not a specially selected model.) After adding the medvram argument, it was successful in loading. The prompt is the same as the previous test photo, with the full-body outfit wildcard applied. a 4:5 aspect ratio of 512*640 Hires 1.5x was successful  CUDA error occurred at hires 1.8x. Tiled VAE started to be applied. Started successfully with default settings. Hires 1.9x was successful  CUDA error occurred at hires 2.0x Lowered the Encoder Tile Size to 1344 Hires 2.0x was successful  Hires 2.5x was successful  The laptop started to lag from hires 2.5x. It seems impossible to do other additional work at the same time. Stable Diffusion WebUI 1.5 version was meaningless to run more than hires 2.5x (I had to increase the denoising to ensu...

Is it too late to selebrate happy new year? Anyway~this year is blue dragon's year So to selebrate this year, I made some pic about blue dragon Trying to add a girl in traditional clothes Hm....Need to seperate girl and dragon. ...Girl! Give me a girl, Stable diffusion! Stupid stable diffusion WebUI Batter to cherry picking. Now here are some final result~ It didn't go well.....Stable diffusion

Pixiv has been suspended, (9th Jan). In fact, my first checkpoint model, plmix3rd was made for uploading to Pixiv. I was very disappointed when I heard the news.   I had checked the real-photo prohibition rules carefully before uploading, and I thought that my model was within the rules So.....I will be crooked from now on chainmail dress Lora Bulk up Lora ,  pantyhose Lora first combination (test) In AD(After Detailer), Demonic_Iconic Lora Chainmail costume Lora is still not working, but I got a good image in the second try Ok. It didn't work in this setting. Attempt to change to the another lora costume( Spiled Latex Dress ) Ok looks good to me For now, let's exclude the lake and beach background wildcard that I made Another clothes Lora( Bondage Harness ) adapted,  increase the number of muscular prompt wieght  Muscular is ok, but didn't show up harness. Adjust Lora weight  I'm settle for this image. I upscaled the images that I liked among the images ...