The imgi blog
Guides on comparing game mods, image edits, and AI upscaler outputs, and sharing before/after sliders online.

Color banding explained: why smooth gradients grow stripes
Why smooth gradients turn into stripes: 8-bit vs 10-bit color, gamma and PQ, dithering, and how to tell banding from compression damage.

Why modern games look blurry: TAA, explained properly
Why modern games look blurry: how TAA works, where the softness and ghosting come from, and what actually helps, from DLAA to supersampling.

Does Removing EXIF Data Reduce Image Quality?
Removing EXIF data doesn't touch a single pixel. The quality loss people blame on it comes from something else the tool did. Here's the real difference.

How to Find Where a Photo Was Taken
Photos can carry the exact GPS of where they were taken. Here's how to find where a photo was taken, and why half the time it just isn't there.

How JPEG compression works (and where the artifacts come from)
How JPEG compression works, step by step: chroma subsampling, DCT blocks, quantization, and why blocking, ringing and mosquito noise look the way they do.

Better lighting in FF7 Rebirth: the Shader Injector mod, before and after
FF7 Rebirth's Shader Injector mod adds contact shadows, reflections and richer lighting. Drag the before and after sliders to see what it changes.

Topaz Gigapixel vs Upscayl: is the free one good enough?
Topaz Gigapixel costs money, Upscayl is free. Here's where the free upscaler holds up, where Topaz still wins, and how to test it on your own image.
How to compare AI upscalers (and pick the one that actually works)
Real-ESRGAN, SwinIR, or Topaz Gigapixel? You can't tell from cherry-picked samples. Upscale your own image with each and drag a slider to compare.

A before and after gallery for your website that people actually use
Most before and after galleries bury the proof. How to build one with draggable sliders that stays fast and actually sells the work.

Website redesign before and after: show the change so clients feel it
A redesign case study works when people can drag between old and new. How to capture matched site screenshots and present the change properly.

Why your images look blurry on Discord and Reddit (and the fix)
Your screenshot was sharp until you posted it. Why chat and social platforms soften images, how to see exactly what got eaten, and how to share losslessly.

Krea 2 vs Stable Diffusion: what four years of open image models actually changed
Krea 2 and Stable Diffusion compared in depth: architecture, training pipelines, licensing, and what four years of open image models changed.