The "Low Resolution" enhancement model of Topaz Photo AI has produced a smudgy result lacking in detail that is comparable to denoised ON1: 60MB PNG compressed to 1.29MB JPG, while the result of a "Natural" model was rich with artifacts of color and detail: 67.9MB PNG compressed to 1.67MB JPG
ON1 Resize AI produced the most noisy result: 67MB PNG compressed to 1.29MB JPG-I manually improved it using their ON1 Photo RAW denoiser module but doing so helped only marginally: 57.5MB PNG compressed to 1MB JPG The free version of letsenhance.io only allowed for 2X upscaling although with PNG format as an option, and the two operations produced a result with noticeable artifacts and smudging of details and colors: 1.29MB JPG Upscayler has produced an almost identical result to, which leads me to believe the latter is a frontend for the former: 1MB JPG produced the most blurry and color-affecting result: 11.5MB JPG Having found a good wallpaper candidate with relevant issues I cropped it to 16:9, added a bit of color depth to it and downscaled it from 4000x2250 to 1280x720 because for whatever reason, most AI image upscalers tested produced much less noisy results after downscaling to <2000px. The other day I had a chance to try out different AI image upscalers/enlargers, namely ON1 Resize AI, Topaz Gigapixel AI, Topaz Photo AI, Upscayl, letsenhance.io, and.
0 kommentar(er)
