Skip to content
View Pfannkuchensack's full-sized avatar
18 followers · 22 following

Achievements

Achievement: Pair Extraordinairex3Achievement: Pull Sharkx2Achievement: Arctic Code Vault Contributor

Achievements

Achievement: Pair Extraordinairex3Achievement: Pull Sharkx2Achievement: Arctic Code Vault Contributor

Highlights

  • Pro

Organizations

@code-with-dot-us @chaostreff-flensburg @invoke-ai

Block or report Pfannkuchensack

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Pfannkuchensack/README.md

Hi, I'm Alexander 👋

Software developer at 54Grad Software in northern Germany. PHP and Node/TypeScript pay the bills; Python and Go are where I spend my evenings.

Most of my open source work orbits InvokeAI, where I contribute upstream — the Z-Image Seed Variance Enhancer and its Linear UI shipped in #8753 — and build the tooling around it that I kept wishing existed.

🎨 InvokeAI ecosystem

Tools

Project What it does Stack
invoke-openai-proxy OpenAI-compatible image API in front of InvokeAI, so Open-WebUI & co. can generate against a local install. Single static binary, embedded admin UI. Go
InvokeAI DB Tool Syncs, restores and cleans the InvokeAI SQLite database against your outputs folder, recovering metadata straight from PNGs. Electron
invokeai-bench Benchmarks txt2img/img2img across SD 1.5 → SDXL, FLUX.1/2, SD3, Z-Image and CogView4, with JSON export and run comparison. Python
invokeai-docker-build CUDA/CPU/ROCm images of my InvokeAI fork, published to GHCR. Docker

Nodes

Project What it does
Qwen3 Prompt Pro Reuses the already-loaded Qwen3 encoder as an LLM to expand prompts, then conditions Z-Image and Flux Klein in one node.
Krea-2 Prompt Weighting Per-token weighting for Krea-2 — (blonde:0), (a red apple:-1) — in ComfyUI, A1111 and compel notation.
Winzige Banane 🍌 Gemini 2.5 Flash / 3 Pro image generation. German for "tiny banana", after Google's Nano Banana codename.
Z-Image Seed Variance Restores seed-to-seed variety in Z-Image-Turbo. Now part of InvokeAI core.
FLUX.2 Hildegard Refiner Tile-based refinement for FLUX.2 Klein: three explicit reference slots (tile, position, global) at full resolution instead of the stock clamped reference path. Ported from 42lux's ComfyUI scheme.
Workflow Package A library of composable workflows that install themselves into InvokeAI.

🕹️ Other things I've built

Game jam entries (GMTK 2021, Global Game Jam 2022, Lost Shapes), a quiz built on InvokeAI's workflow engine, and hardware bits and pieces from Chaostreff Flensburg/Chaos Computer Club Flensburg.

🎩

On GitHub since 2012 · Arctic Code Vault contributor

Pinned Loading

  1. invoke-ai/InvokeAI invoke-ai/InvokeAI Public

    Invoke is a leading creative engine for Stable Diffusion models, empowering professionals, artists, and enthusiasts to generate and create visual media using the latest AI-driven technologies. The …

    Python 28k 2.9k

  2. sqlite_invokeai_db_tool sqlite_invokeai_db_tool Public

    Desktop tool to sync, repair and clean the InvokeAI database against your outputs folder, recovering metadata from PNGs

    JavaScript 4

  3. openaiapi2invokeai-go openaiapi2invokeai-go Public

    OpenAI-compatible image API proxy for InvokeAI — a single static Go binary with an embedded admin UI

    Go 2 1

  4. flux2_hildegard_refiner flux2_hildegard_refiner Public

    InvokeAI nodes for tile-based FLUX.2 Klein refinement with explicit tile, position and global reference latents (Hildegard scheme)

    Python

  5. invokenode_qwen3_prompt_pro invokenode_qwen3_prompt_pro Public

    InvokeAI node that reuses the already-loaded Qwen3 encoder as an LLM to enhance prompts, conditioning Z-Image and Flux Klein in one step

    Python

  6. krea2_prompt_weighting krea2_prompt_weighting Public

    Per-token prompt weighting for Krea-2 in InvokeAI, accepting ComfyUI, A1111 and compel notation

    Python