Skip to main content

Indie game storeFree gamesFun gamesHorror games
Game developmentAssetsComics
SalesBundles
Jobs
TagsGame Engines

Cuong Tran | Human Pixel Art Fix

14
Posts
1
Topics
A member registered 6 days ago · View creator page →

Recent community posts

That combo sounds like the best of both: the glow says "something's here" from across the room, and the outline confirms what you can actually use up close. One thing to watch: keep the glow a step less saturated than enemy/hit effects so it never reads as danger. Looking forward to the next build!

Ah, I missed that, the pulse must be getting lost against the floor at that zoom. A darker silhouette or a 1px dark outline under the glow usually helps a lot, since contrast in value reads before color does. Looking forward to the next build!

From the screenshots, the floor is big and low-contrast and the interactables share its brown tones, so they melt in. A 1px bright rim or a slow 2-3 frame glow pulse on doors, portals and the shop, in a color nothing else on the map uses, would make them readable before the "Press E" prompt even shows up.

Just checked the v0.12.2 notes, that was fast! Taller banners and redrawn icons are exactly what I hoped for. I'll play a longer session this weekend and share notes on pacing and mobile.

Some first-impression notes from the art side, based on the screenshots: the terminal/neon UI is clean and very readable, and the "What next" panel is a great way to guide the first 15 minutes. The pixel banner at the top of each section is a nice touch, but at that strip height the character and props are tiny next to the big crisp UI text, so the pixel art barely gets to set the mood. Giving the scene more height, or showing it at a clean 2x/3x scale, would help a lot. Same for the skill icons in the sidebar: one size step up would make them easier to tell apart on phones. Will report back after a longer session.

Sent you a friend request on Discord (cuongtran)

Thanks for the detailed list, that helps a lot! Before I put numbers on it, a couple of quick questions: how many directions do you need per monster for battle, 1 (front-facing), 2 (front + back), or 4? And roughly how many frames per animation are you imagining? Both change the scope quite a bit. 
Would you be up for continuing on Discord? It'd be easier to share my animation process and a short demo there so you can see if the style and workflow fit before we talk pricing. I'm cuongtran on Discord, feel free to add me.

Hi Kaloth! MonsterMaster sounds fun, DQM-style synthesis is a great hook. I'm a pixel artist focused on making sprite sets consistent: unified scale, palette and outline rules, plus animation (idle/attack/hit/death). Since you mentioned using some AI already, I also clean up AI-generated sprites into proper grid-aligned pixel art, which can be an affordable way to make a mixed set look like one game. Happy to start with a small paid milestone, e.g. one monster with idle + attack. What sprite size are you targeting for battle monsters? Portfolio: https://cuongpixelfix.itch.io

Looking at the screenshots with the graphics bugs question in mind: the overworld already has a cozy read. The tree canopies are a bit softer than the buildings, so some leaf clusters sit off the same pixel grid as the roofs. A pass to snap those edges, plus a consistent 1px outline on the player and NPC sprites, would make the characters pop against the grass. The bee on the battle screen is charming; tightening the outline and cleaning up stray anti-aliased pixels on the fuzz would match the cleaner building style.

Exactly. I usually flip through frames side by side and check the face/hands per direction. That's where most of the cleanup time goes.

AI “pixel art” often looks fine until you zoom to 1x. The usual problems are not dramatic: the grid isn’t even, there are dozens of almost-identical colors, and a 1-pixel fringe of anti-aliasing sits around the silhouette.

One check I use on every sprite: put it on both a light background and a dark one, at actual size. If you see a halo, leftover anti-alias, or a cluster of orphan pixels, it will show up the same way in-engine. After that I lock a small shared palette, even out the outline width, and, for animations, onion-skin the sheet so the feet stay on one pivot.

I’m Cuong (Human Pixel Art Fix / cuongpixelfix). I do this cleanup by hand and can send a short timelapse plus the layered .aseprite file, which is also useful if you need process proof for a devlog or Steam disclosure.

If anyone wants, I’ll clean up one idle sprite for free this week — one character, 32x32 or 64x64, no animation sheet. Reply here with the PNG and the target size, and I’ll send back a before/after plus the .aseprite file. I’ll only take a few so I can do them properly.

This is a very solid pipeline, especially doing the cutout at the small grid size — fringes are almost invisible on the 512px version. For a shared palette across a pack, I usually build it from 3–4 hero sprites, then force every later piece onto that same ramp: one outline color, 2–3 shadow steps, and I merge near-identical hues instead of keeping all of them. I also test tiles on a repeating map and sprites on both light and dark ground, because a halo or a one-off color only shows up then. Thanks for writing the steps out so clearly.

Agree that dishonest 'no AI' tags are the part that breaks trust, not disclosure itself. Process recordings and layered source files are becoming the practical proof a buyer or Steam page can actually check, worth attaching even when the work is fully hand-drawn.

This is a really useful first step. After the snap I still usually check animation frames for jitter and whether the character looks like the same person from every direction — the grid tool can't catch that.