Credits to George Bailey for a template spritesheet (licensed under CC-BY 4.0), which was edited and used to make the dataset for the LoRA

opengameart.org/content/16x1...

svntax-dev/pixel_spritesheet_4walk_small_lora_v1 · Hugging Face We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Links to the LoRA:

Hugging Face - huggingface.co/svntax-dev/p...

Civitai - civitai.com/models/23563...

I trained a pixel art spritesheet LoRA on FLUX.2 Klein base 4B. For a first attempt using small sprites, I think it's pretty good.

First 3 rows are text-to-image. Last row uses edit workflow with image references.

Generated pixel art spritesheet of four 64 x 64 pixel frames in a 2 by 2 grid of a light blue haired swordswoman with armor. The poses are from left to right, top to bottom, standing idle, attacking pose, hurt pose, kneeling pose.

Generated pixel art spritesheet of four 64 x 64 pixel frames in a 2 by 2 grid of a light blue haired swordswoman with armor. The poses are from left to right, top to bottom, standing arms crossed pose, one arm pointing right pose, both arms stretched outwards pose, hands covering mouth pose.

Generated pixel art spritesheet of four 64 x 64 pixel frames in a 2 by 2 grid of a red bird man wearing leather shoulder armor and a dark purple shirt. The poses are from left to right, top to bottom, standing idle, attacking pose, hurt pose, kneeling pose.

Generated pixel art spritesheet of four 64 x 64 pixel frames in a 2 by 2 grid of a red bird man wearing leather shoulder armor and a dark purple shirt. The poses are from left to right, top to bottom, standing arms crossed pose, one arm pointing right pose, both arms stretched outwards pose, hands covering mouth pose.

Sadly, the actual spritesheets from this LoRA are just too messy. Prompt adherence is pretty good, but I think the dataset and the fact that Z-Image Turbo is distilled make the pixels noisy and inconsistent. Or maybe 64x64 sprites are just too large and more prone to issues.

Generated pixel art portrait at 128 by 128 resolution of a light blue haired swordswoman with armor.

Generated pixel art portrait at 64 by 64 resolution of a light blue haired swordswoman with armor.

Generated pixel art portrait at 128 by 128 resolution of a red bird man wearing leather shoulder armor and a dark purple shirt.

Generated pixel art portrait at 64 by 64 resolution of a red bird man wearing leather shoulder armor and a dark purple shirt.

In the process of training pixel art spritesheet LoRAs, I've noticed that Z-Image Turbo is capable of more general images like portraits, even though the dataset is purely spritesheets.

Sample images at 64x64 and 128x128 (same post-processing steps for all: k-centroid downscale, palettize).

Pixel art portrait of a witch

Pixel art portrait of a knight

Pixel art portrait of a werewolf

I trained a Z-Image pixel art LoRA just for testing purposes. Even though it's a distilled model (and my dataset is rough), results are not too bad with a little post-processing.

The Steam Frame looks really cool. It's another step towards a fully open source VR headset.

I made a ComfyUI custom node for using the Retro Diffusion API in your workflows, and it's now published to the Comfy Registry.

Source code - github.com/svntax/ComfyUI-RetroDiff...

Welcome to the NEAR Legion! | NEAR Legion Guide

If you're wondering what's NEAR Legion:

It's a community with people and teams focused on open, user-owned AI on NEAR. There's cool info there on what's being worked on, and I'm aiming to shine a spotlight on the gamedev side of things.

More info here - docs.nearlegion.com

Play the closed alpha here - svntax.itch.io/the-near-mines

Testing is open to NEAR Legion members. Password is in the group.

Not a member? You can join with my referral link - app.nearlegion.com?referral=5f2...

I've been working on a Minesweeper-like game using a Godot + NEAR Protocol SDK I made years ago, now up-to-date with support for Intear wallet.

You play as a dwarf digging for gold & gems while avoiding traps.

There was a free LoRA training week on Hugging Face recently (LoRA Frenzi), so I trained a pixel portrait LoRA for Qwen Image.

huggingface.co/svntax-dev/pixel_portrai...

I wrote a blog post about my (unfinished) VR demo for #js13k using PlayCanvas. There were some tough but interesting limits to using it while keeping the game under 13 KB.

blog.svntax.com/playcanvas-and-the-js13k...

It's been about a month since I started using kanata (https://github.com/jtroo/kanata for remapping keys. It's so helpful not having to move my hand to the arrow keys as often when writing (using caps lock to switch layers).

Hoping to make some time this month for #js13k

GitHub - svntax/smear-spritesheet-generator: A tool for generating spritesheets with smear effects. A tool for generating spritesheets with smear effects. - svntax/smear-spritesheet-generator

Now public and open source - github.com/svntax/smear...

You can set a subregion of a sprite so that smear effects are applied to that area only.

I tried making a procedural animation tool that generates smear frames for any image. It's bare-bones, but working.

RPG prototype with generated characters using Retro Diffusion's API, set up as a summoning mechanic.

In the process of making an RPG prototype with smart NPCs using the new Retro Diffusion API updates, I found out about Dave Mark's Infinite Axis Utility System (IAUS). It looks far more flexible than GOAP for dynamic planning.

I wish I heard of Excalidraw earlier. I've been seeing it used for system design/architecture diagrams lately, which looks really helpful. Going to figure out how to self-host it now.

Finished setting up Solidtime for tracking time for my projects. I've used Clockify for a long time, but I wanted a self-hosted alternative, and so far it's been working great.

I got it to work after watching the video, but you're right. I'm also getting issues with self-signed certs when uploading images. At least posting text works, so thanks again.

Testing Postiz again, finally got the user signup working. Now to see if I can post this on multiple platforms.

Thanks for the video. I'm using Traefik for the reverse proxy, but I'll check this out anyway. Maybe I'm missing something in my configuration somewhere.

So far I've tried self-hosting Postiz and Mixpost. Postiz I got running up until user signup, which fails for some reason, and no helpful errors in the logs. Mixpost worked great, but the Lite version doesn't support all the platforms I need.

Currently figuring out how to set up a posts scheduling tool on a local server.