LOADING DATASHEET
LOADING DATASHEET
by Lodestones
RANKED #76 OF 101 IMAGES · OVERALL #388 OF 493 RANKED · VIBE SCORE 5.5 · 44 VOICES ·
COPY BADGE pastes a README snippet with the rank SVG.
Chroma1-HD on Hugging Face (text to image). 68,678 downloads. Open weights for local or hosted use.
14 mentions
1 mentions
26 weeks · 20 voices in this window
HOLDING UP
Recent vibe is steady. No clear fade in the crowd.
Open weights. You can use a hosted service, or download it and run it yourself, free.
Open weights: download Chroma1 HD once and it's yours. No subscription, no rate limits, works offline.
8.9B parameters, an FP8/Q4 checkpoint is about 5 to 9 GB while BF16 is ~16.6 GB
NVIDIA GeForce RTX 3060 12GB
12 GB VRAM
~$260 USEDGGUF Q4 or SDNQ 4-bit with CPU offloaded T5 encoder at 1024x1024
NVIDIA GeForce RTX 4070 Ti Super 16GB
16 GB VRAM
~$750 NEWFP8 scaled checkpoint with FP8 T5 encoder at 1024x1024
NVIDIA GeForce RTX 4090 24GB
24 GB VRAM
~$1650 USEDFull BF16 / FP16 precision transformer with T5-XXL text encoder
The three cards above are researched picks. Search the machine you actually have — we only say yes if it should reply at a usable speed, not just load the weights and crawl.
HARDWARE PICKS & PRICES ARE RESEARCHED FROM THE LIVE WEB AND REFRESHED AUTOMATICALLY. TREAT THEM AS BALLPARK, NOT GOSPEL.
Everyday jobs on each of those rigs: how long Chroma1 HD takes, and how good the crowd says it is at that kind of work.
| THE JOB | BARE MINIMUMNVIDIA GeForce RTX 3060 12GB | SWEET SPOTNVIDIA GeForce RTX 4070 Ti Super 16GB | FULL POWERNVIDIA GeForce RTX 4090 24GB |
|---|---|---|---|
| Make an imageone picture at normal settings | ~45 S★★★★★ | ~18 S★★★★★ | ~8 S★★★★★ |
STARS ARE THE COMMUNITY SCORE FOR THAT KIND OF WORK, DOCKED FOR HOW SQUASHED THE WEIGHTS GET AT EACH TIER. TIMES ARE COMPUTED FROM RESEARCHED SPEEDS. BALLPARK, NOT GOSPEL.
Not enough discussion about Chroma1 HD yet to call it. We found 44 posts but people did not say much either way.
30 POSTS · 14 COMMENTS · REDDIT · GITHUB
Other models the crowd has fully reviewed, starting with images models like this one.
Update: Chroma Project training is finished! The models are now released.
Pushing the limits of Chroma1-HD
Cross-Image Try-On Flux Kontext_v0.2
Chroma1-HD + r64-flash-huen-lora + lenovo-ultrareal-lora (CFG = 1).
A request to anyone training new models: please let this composition die
Comic, oil painting, 3D and a drawing style LoRAs for Chroma1-HD
Chroma v48 to Chroma1-Base
(5) The same message applies to several models: Chroma, Z image, Klein, Ernie, Qwen 2512
Character Workflow: Chroma1-HD + Flux.2 Dev + Wan 2.2 + LTX 2.3
(3) The same message applies to several models: Chroma, Z image, Klein, Ernie, Midjourney
Chroma1-HD Character Transfer with Flux.2 Dev
This is an open source project (MIT license, no monetization) built specifically around local LLMs I'm not a PC expert at all and I'm completely new to the whole LLM and local LLM scene. Before running models locally I used AI Dungeon — I really enjoyed it at…