Mindket

Marigold V2 klein Normal to Height

Newversion 1 · updated

Overview

Turns a tangent-space normal map into a height map in one transformer step, using Marigold V2 on FLUX.2 klein 4B. A Gradio app for texture and 3D material work.

Marigold V2 klein Normal to Height is a kit in the Productivity category, listed by Mindket. It is free. This is version 1, last updated on . Download it on this page, a zip of its files. Nothing in it runs by itself.

At a glance

You need
  • Nothing at all to use the online Space
  • Python 3.12 and pip, for a local copy
  • The Python packages listed in requirements.txt
  • A read token for a private model repository, set as the HF_TOKEN secret
Runs in
  • A web browser, on the Space's page at Hugging Face
  • A local copy on your own computer, with Python 3.12 installed
Start with
pip install -U huggingface_hub
hf download NightRaven109/marigold-v2-klein-normal-to-height --repo-type space --local-dir marigold-v2-klein-normal-to-height
cd marigold-v2-klein-normal-to-height

About

What it is

This is a Gradio app that makes a height map from a tangent-space normal map. The model is a Marigold V2 variant built on FLUX.2 klein 4B and trained further with Megascans and MatSynth data. One transformer pass turns the normal map into the height map. It runs on zero-a10g hardware at Hugging Face.

Who it suits

Texture and 3D material artists who have a normal map and need a matching height map, and developers who want to see how the model is wrapped in an app.

IndustryWho uses itHow they use itExample
GamesEnvironment artistsGet a height map to go with a materialConvert the normal map of a stone wall
SoftwareDevelopersRead app.py and tiling.py to see how images are processedReuse the Gradio layout
ResearchGraphics researchersTry a normal-to-height modelRun a few maps and inspect the results

Skip it if you start from photos or color maps, since the input is a normal map.

Where to use it

  • A web browser, on the Space's page at Hugging Face
  • A local copy on your own computer, with Python 3.12 installed
Use caseWhoWhat they get back
Make a height mapTexture artistsA height map from your normal map
Try the modelResearchersA way to test the normal-to-height fine-tune

How to set it up

  • Nothing at all to use the online Space
  • Python 3.12 and pip, for a local copy
  • The Python packages listed in requirements.txt
  • A read token for a private model repository, set as the HF_TOKEN secret
  1. Open the Space's page at Hugging Face to use it online.
  2. For a local copy, install the Hugging Face hub package and download the Space.
  3. Move into the new folder and install Gradio at the pinned version.
  4. Install the packages from requirements.txt.
  5. Make sure HF_TOKEN holds a read token that can reach the model repository, then start the app.
pip install -U huggingface_hub
hf download NightRaven109/marigold-v2-klein-normal-to-height --repo-type space --local-dir marigold-v2-klein-normal-to-height
cd marigold-v2-klein-normal-to-height
pip install gradio==6.24.0
pip install -r requirements.txt
python app.py

The weights sit in a private model repository, which the token must be able to read. A setting called MODEL_REPO can point to a different repository id.

How to use it

Online, open the Space and give it a normal map through the form on its page. On your own machine, run the app file and open the page Gradio serves in your browser. The page lists no other commands.

ToolWhat it does
python app.pyStarts the Gradio app on your computer
HF_TOKENSecret that holds a read token for the model repository
MODEL_REPOOverrides the repository id the weights come from

Example

You upload the normal map of a brick wall and run the conversion. The app returns a height map of the wall in a single step, which you can save for use next to your other texture files.

For newcomers and experts

If you are new, a normal map stores surface direction, while a height map stores how high each point sits. Open the Space online and try one small normal map first.

If you know the field, this is a fine-tune of Marigold V2 on the FLUX.2 klein 4B base, trained with Megascans and MatSynth, and inference needs only one transformer pass. A tiling.py file sits beside app.py, but the page does not explain it.

  1. Upload a tangent-space normal map
  2. Marigold V2 on FLUX.2 klein 4B runs once
  3. A height map is predicted
  4. You download the height map

Good to know

  • License: apache-2
  • Cost to run: none named, though a Hugging Face read token is needed for the private weights
  • Data it reaches: the normal map you upload, and the model weights read from a private repository
  • Not for: starting from photos or color maps, since the input is a tangent-space normal map

README and files

README

Marigold V2 klein Normal to Height

Turns a tangent-space normal map into a height map in one transformer step, using Marigold V2 on FLUX.2 klein 4B. A Gradio app for texture and 3D material work.

Made by NightRaven109 on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: This is a Gradio app that makes a height map from a tangent-space normal map. The model is a Marigold V2 variant built on FLUX.2 klein 4B and trained further with Megascans and MatSynth data. One transformer pass turns the normal map into the height map. It runs on zero-a10g hardware at Hugging Face.

Who it suits: Texture and 3D material artists who have a normal map and need a matching height map, and developers who want to see how the model is wrapped in an app.

Industry | Who uses it | How they use it | Example
Games | Environment artists | Get a height map to go with a material | Convert the normal map of a stone wall
Software | Developers | Read app.py and tiling.py to see how images are processed | Reuse the Gradio layout
Research | Graphics researchers | Try a normal-to-height model | Run a few maps and inspect the results

Skip it if you start from photos or color maps, since the input is a normal map.

Where to use it:

  • A web browser, on the Space's page at Hugging Face
  • A local copy on your own computer, with Python 3.12 installed
Use case | Who | What they get back
Make a height map | Texture artists | A height map from your normal map
Try the model | Researchers | A way to test the normal-to-height fine-tune

How to set it up:

  • Nothing at all to use the online Space
  • Python 3.12 and pip, for a local copy
  • The Python packages listed in requirements.txt
  • A read token for a private model repository, set as the HF_TOKEN secret
  1. Open the Space's page at Hugging Face to use it online.
  2. For a local copy, install the Hugging Face hub package and download the Space.
  3. Move into the new folder and install Gradio at the pinned version.
  4. Install the packages from requirements.txt.
  5. Make sure HF_TOKEN holds a read token that can reach the model repository, then start the app.
pip install -U huggingface_hub
hf download NightRaven109/marigold-v2-klein-normal-to-height --repo-type space --local-dir marigold-v2-klein-normal-to-height
cd marigold-v2-klein-normal-to-height
pip install gradio==6.24.0
pip install -r requirements.txt
python app.py

The weights sit in a private model repository, which the token must be able to read. A setting called MODEL_REPO can point to a different repository id.

How to use it: Online, open the Space and give it a normal map through the form on its page. On your own machine, run the app file and open the page Gradio serves in your browser. The page lists no other commands.

Tool | What it does
python app.py | Starts the Gradio app on your computer
HF_TOKEN | Secret that holds a read token for the model repository
MODEL_REPO | Overrides the repository id the weights come from

Example: You upload the normal map of a brick wall and run the conversion. The app returns a height map of the wall in a single step, which you can save for use next to your other texture files.

For newcomers and experts: If you are new, a normal map stores surface direction, while a height map stores how high each point sits. Open the Space online and try one small normal map first.

If you know the field, this is a fine-tune of Marigold V2 on the FLUX.2 klein 4B base, trained with Megascans and MatSynth, and inference needs only one transformer pass. A tiling.py file sits beside app.py, but the page does not explain it.

Upload a tangent-space normal map
Marigold V2 on FLUX.2 klein 4B runs once
A height map is predicted
You download the height map

Good to know:

  • License: apache-2
  • Cost to run: none named, though a Hugging Face read token is needed for the private weights
  • Data it reaches: the normal map you upload, and the model weights read from a private repository
  • Not for: starting from photos or color maps, since the input is a tangent-space normal map

On Hugging Face, 13 November 2026

  • Likes: 1
  • Built with: gradio
  • License: apache-2.0

Links

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.

Files
README.md  5 KB

1 file, 5 KB, downloaded as a zip.

What’s new

  • Version 1 · 2026-11-12

Compare with similar

Listed Hugging Face Spaces in the same category that share the most tags with this one.

ProductLikesBuilt withLicenseLast change
Marigold V2 klein Normal to Height This page1gradioapache-2.0
MusicGenShares gradio5,116gradiocc-by-nc-4.0
Kokoro TTSShares gradio3,556gradioapache-2.0
Background RemovalShares gradio2,988gradiomit
Omni Image EditorShares gradio2,922gradiomit
Diffusers Image OutpaintShares gradio2,565gradioapache-2.0
Numbers as Mindket last read them, on 2026-11-12.

Reviews

No ratings yet

Write a review

No reviews yet.

More from Mindket

See all from Mindket

anaksor-ocr

Sokheng · Kit

Reads Khmer and English text from documents, receipts and photos of signs, using a detector for text lines and a recogniser for each line. Suits developers who…

New

Qwen2.5-7B-Instruct as GGUF files in eight quantization sizes, to chat with locally through llama.cpp. For people who want a capable 7.6B model on their own co…

New

ACE-Step 1.5 XL SFT

ACE-Step · Kit

Generates music from text with a 4B diffusion decoder, 50 steps and guidance control, aiming for higher audio quality than the base. For creators with a GPU of…

New

Similar in Productivity

See all in Productivity

Guizang PPT Skill

op7418 · Skill

Skill that makes single-file HTML slide decks, slide illustrations and platform covers in two visual styles, with a presenter mode for rehearsal. For speakers…

New29.8k

Frontend Slides

zarazhangrui · Skill

Builds HTML presentations from scratch or from a PowerPoint file and shows visual style previews to pick from, for people who want web slides with no CSS or Ja…

New32.3k

IPTV

iptv-org · Kit

Collects publicly available live TV channels from many countries into playlists that any video player with live streaming can open. Made for viewers and for de…

New144k