jc-builds commited on
Commit
722a672
·
verified ·
1 Parent(s): ec61d63

KilnImage 0.1.0: initial Chroma1-HD iOS bundle

Browse files
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ Chroma1-HD-Q4_K_S.gguf filter=lfs diff=lfs merge=lfs -text
Chroma1-HD-Q4_K_S.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4443db48850a45bb7f163a0582ea0e9f9d449db1aa56632c8572515e8e83acc8
3
+ size 5432053920
README.md ADDED
@@ -0,0 +1,80 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ language: en
4
+ pipeline_tag: text-to-image
5
+ tags:
6
+ - text-to-image
7
+ - diffusion
8
+ - chroma
9
+ - flux
10
+ - gguf
11
+ - quantized
12
+ - on-device
13
+ - ios
14
+ - mobile
15
+ - apple-silicon
16
+ base_model: lodestones/Chroma1-HD
17
+ ---
18
+
19
+ # Chroma1-HD — iOS bundle
20
+
21
+ <p align="center">
22
+ <a href="https://github.com/haplollc/Mirage"><img alt="Mirage" src="https://img.shields.io/badge/Runs%20on-Mirage-orange" /></a>
23
+ <a href="https://huggingface.co/lodestones/Chroma1-HD"><img alt="Upstream" src="https://img.shields.io/badge/Upstream-lodestones%2FChroma1--HD-blue" /></a>
24
+ <img alt="License" src="https://img.shields.io/badge/license-Apache--2.0-lightgrey" />
25
+ <img alt="Params" src="https://img.shields.io/badge/params-8.9B-purple" />
26
+ </p>
27
+
28
+ A mobile-friendly bundle of **Chroma1-HD** (8.9B Flux-derived DiT) packaged with T5-XXL + Flux VAE, sized for on-device inference via [**Mirage**](https://github.com/haplollc/Mirage).
29
+
30
+ Chroma1-HD is the high-resolution flagship variant from [lodestones/Chroma1-HD](https://huggingface.co/lodestones/Chroma1-HD) — fully open weights built on FLUX.1's architecture with significant retraining for general-purpose generation. Excellent at photorealism, illustration, and concept work.
31
+
32
+ ## What's inside
33
+
34
+ | File | Role | Size |
35
+ |---|---|---|
36
+ | [`Chroma1-HD-Q4_K_S.gguf`](./Chroma1-HD-Q4_K_S.gguf) | Diffusion transformer — Flux-arch, 8.9B params, Q4_K_S | 5.1 GB |
37
+ | [`t5xxl_fp16.safetensors`](./t5xxl_fp16.safetensors) | Text encoder (T5-XXL, fp16) | 9.1 GB |
38
+ | [`ae.safetensors`](./ae.safetensors) | VAE (from FLUX.1) | 320 MB |
39
+
40
+ Total bundle: **~14.5 GB**. Total GPU residency: ~15-16 GB. **iPhone 17 Pro / Air (12 GB) or M-series Mac required.** Older iPhones cannot hold this model.
41
+
42
+ ## Quick start (Mirage)
43
+
44
+ ```swift
45
+ import Mirage
46
+
47
+ let docs = FileManager.default.urls(for: .documentDirectory, in: .userDomainMask)[0]
48
+
49
+ let engine = try Engine(models: ModelFiles(
50
+ diffusionModel: docs.appendingPathComponent("Chroma1-HD-Q4_K_S.gguf"),
51
+ vae: docs.appendingPathComponent("ae.safetensors"),
52
+ textEncoder: docs.appendingPathComponent("t5xxl_fp16.safetensors")
53
+ ))
54
+
55
+ let image = try await engine.generate(.init(
56
+ prompt: "a Hopper-style oil painting of an empty diner at dawn, soft warm light",
57
+ width: 1024, height: 1024,
58
+ steps: 28,
59
+ cfgScale: 4.0 // Chroma's recommended CFG
60
+ ))
61
+ ```
62
+
63
+ ## Provenance
64
+
65
+ | Component | Upstream | License |
66
+ |---|---|---|
67
+ | Diffusion transformer | [lodestones/Chroma1-HD](https://huggingface.co/lodestones/Chroma1-HD) | Apache 2.0 |
68
+ | GGUF conversion | [silveroxides/Chroma1-HD-GGUF](https://huggingface.co/silveroxides/Chroma1-HD-GGUF) | Apache 2.0 |
69
+ | Text encoder | [comfyanonymous/flux_text_encoders](https://huggingface.co/comfyanonymous/flux_text_encoders) (`t5xxl_fp16.safetensors`) | Apache 2.0 |
70
+ | VAE | [ffxvs/vae-flux](https://huggingface.co/ffxvs/vae-flux) (re-host of FLUX.1's `ae.safetensors`) | FLUX-1-dev-non-commercial |
71
+
72
+ ## Sizing
73
+
74
+ - **iPhone 17 Pro / Air (12 GB RAM)** — works but tight; expect ~8 min per 1024² image.
75
+ - **iPhone 16 Pro (8 GB RAM)** — does NOT fit. The text encoder alone is 9 GB. Use [`Z-Image-Turbo-iOS`](https://huggingface.co/jc-builds/Z-Image-Turbo-iOS) instead.
76
+ - **M2 / M3 Mac** — comfortable, ~4-6 min per 1024² image at 28 steps.
77
+
78
+ ## Built by
79
+
80
+ [Haplo](https://haplo.app) · [Mirage on GitHub](https://github.com/haplollc/Mirage)
ae.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:afc8e28272cd15db3919bacdb6918ce9c1ed22e96cb12c4d5ed0fba823529e38
3
+ size 335304388
t5xxl_fp16.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e480b09fae049a72d2a8c5fbccb8d3e92febeb233bbe9dfe7256958a9167635
3
+ size 9787841024