doombatz it is so much fun to generate and many things out of the Box, without a Lora, that i like very much and I have many fetishes :)

And it is the very first one what understands what a hogtie is :) very nice

1280 x 1536 works very well, i like this size

Gorean You won't get weird lines at the edges since both numbers are dividable by 32, but the total number of pixels is 87.5% higher than 1024x1024. So depending on the model, you will more often encounter deformations and errors in your images. Therefore I rather use tiled upscaling (with each tile still being 1 megapixel in size) to increase the resolution above the usual 1 megapixel. In principle you can increase the resolution with tiled upscaling to any value you want, but the computations will take longer and longer. I usually upscale my generations to about 9 megapixel, which for me is a good balance of quality and time.

lizardon1024 I know what you mean and normaly i do so, but this model dont need it, you can use out of the Box 1280x1536 without more deformations. This is incredible and one of the many reasons why I like this checkpoint so much. I tryed to use bigger, there it starts with 3 Legs, 2 Heads, 4 Hands etc. But 1280x1536 with this checkpoint works very well for me. All my latest pictures you see below are 1280 x 1536 without Lora (except petplay and ball gag) or upscale, only facedetailer but i didnt test without facedetailer. Maybe it work without to.

Gorean It doesn't have to lead to deformations but in your latest examples there are clearly images that have deformations which are probably caused by your increased resolution like this and this. The torso is elongated and the hands and feet are weirdly deformed. This might be in part caused by the LoRA, but torso elongation is the most common problem of using a resolution that is too high.

Thank you, this is really useful information!