True 4K Image generation BUILT FOR LOW VRAM! <3

a simple upscaling workflow for Flux Dev using the newest upscaling method with the DyPE node!

v2.0 with the 2nd pass using SRPO model for refiner values are standard in the workflow provided.

changes you can make for variation, quality, or speed are:

--steps on SECOND ksampler - 5-20 steps MAX (do not exceed 20 steps of refinement if your base image on first pass was 20 or more. morphing and degregation occurs after too many steps.)

--denoise on SECOND ksampler - .1 - .8 (.8 is alot of refinement and will essentially change the image entirely based on the SRPO models dataset. start with .15 and work your way up accordingly)

--Seed - self explanatory.

a short description for how dype works:

DyPE is a novel, training-free method that allows pre-trained diffusion transformers like FLUX to generate images at resolutions far beyond their training data, with no additional sampling cost. It works by taking advantage of the spectral progression inherent to the diffusion process. By dynamically adjusting the model's positional encodings at each step, DyPE matches their frequency spectrum with the current stage of the generative process—focusing on low-frequency structures early on and resolving high-frequency details in later steps. This prevents the repeating artifacts and structural degradation typically seen when pushing models beyond their native resolution.

generation times are slightly longer than regular upscale methods like realESERGAN but results seem to be on par if not better in certain aspects. that being said, this node is a lot less resource intensive, requiring nothing extra to keep the model running without errors.