Is there any AI upsampler that is 100% true to the low-res image?

There is a way to guarantee that an upsampled image is accurate to the low-res image: when you downsample it again, it is pixel-perfect the same. There are many possible images that have this property, including some that just look blurry. But every AI upsampler I've tried that adds in details does NOT have this property. It makes at least minor changes. Is there any I can use that I will be sure DOES have this property? I know it would have to be differently trained than they usually are. That's what I'm asking for.

https://redd.it/1px0rd7
@rStableDiffusion

Will there be a quantization of TRELLIS2, or low vram workflows for it? Did anyone make it work under 16GB of VRAM?



https://redd.it/1px8q8r
@rStableDiffusion

Wan 2.2 More Consistent Multipart Video Generation via FreeLong - ComfyUI Node
https://www.youtube.com/watch?v=wZgoklsVplc

https://redd.it/1px9t51
@rStableDiffusion

The LoRAs just keep coming! This time it's an exaggerated impasto/textured painting style.

https://redd.it/1px705k
@rStableDiffusion

Invoke is revived! Crafted a detailed character card by compositing around 65 Z-Image Turbo layers.
https://redd.it/1pxesor
@rStableDiffusion

[SD1.5] This image was entirely generated by AI, not human-prompted (explanation in the comments)
https://i.imgur.com/E0bv2qo.png

https://redd.it/1pxg7n7
@rStableDiffusion

(ComfyUI with 5090) Free resources used to generate infinitely long 2K@36fps videos w/LoRAs

I wanna share what is possible to achieve on a single RTX 5090 in ComfyUI. In theory it's possible to generate infinitely long coherent 2k videos at 32fps with custom LoRAs with prompts on any timestamps. My 50-sec video was crisp and beautiful motions and had no distortion or blur and also character consistency throughout the video with my start image.

Stats on a 50-sec generation:

SVI 2.0 Pro (WAN 2.2 A14B I2V):

50-second video (765 frames): Generate 1280x720 = 1620 secs [SageAttn2 and Torch Compile w/latest lightx2v\]

SeedVR2 v2.5.24 (ema_7b_fp16):

50-second video (765 frames): Upscale 1280x720 to 2560x1440 = 1984 secs [SageAttn2 and Triton - Torch Compile could be used here as well, I just forgot\]

Rife VFI (rife49):

50-second video (1530 frames): Frame Interpolation 16fps to 32fps = 450 secs

Video Combine:

50-second video (1530 frames): Combine frames = 313 secs

Total = 4367 secs (72 mins) for a crisp and beautiful (no slowmotion) 2560x1440 video with 36 fps.

I might drop a video later in a new post, and if enough people would like a ComfyUI workflow, I will share it.

https://preview.redd.it/0deireppyw9g1.png?width=1058&format=png&auto=webp&s=e65f460de8343b620aca5c2764b38e3a054ce5b8



https://redd.it/1pxn75c
@rStableDiffusion

Render in SD - Molded in Blender - Initially drawn by hand

https://redd.it/1pxotth
@rStableDiffusion