Instructions to use Alex995647/loras-minimax-h3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use Alex995647/loras-minimax-h3 with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("MiniMaxAI/MiniMax-H3", dtype=torch.bfloat16, device_map="cuda") pipe.load_lora_weights("Alex995647/loras-minimax-h3") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
Add Minimax H3 Photorealistic Image Generator (LORA + Workflow)
Browse files- .gitattributes +6 -0
- minimax-h3-photorealistic-image-generator-lora-workflow/example_images/01_poster.jpg +0 -0
- minimax-h3-photorealistic-image-generator-lora-workflow/example_images/01_video.mp4 +3 -0
- minimax-h3-photorealistic-image-generator-lora-workflow/example_images/02_image.jpeg +3 -0
- minimax-h3-photorealistic-image-generator-lora-workflow/example_images/03_image.jpeg +3 -0
- minimax-h3-photorealistic-image-generator-lora-workflow/example_images/04_image.jpeg +3 -0
- minimax-h3-photorealistic-image-generator-lora-workflow/example_images/05_image.jpeg +3 -0
- minimax-h3-photorealistic-image-generator-lora-workflow/example_images/06_image.jpeg +3 -0
- minimax-h3-photorealistic-image-generator-lora-workflow/h3_photoreal_ph0t0r34l_1024-step00003780.safetensors +3 -0
- minimax-h3-photorealistic-image-generator-lora-workflow/info.txt +253 -0
- minimax-h3-photorealistic-image-generator-lora-workflow/metadata.json +139 -0
.gitattributes
CHANGED
|
@@ -124,3 +124,9 @@ drone-shot/example_images/05_poster.jpg filter=lfs diff=lfs merge=lfs -text
|
|
| 124 |
drone-shot/example_images/05_video.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 125 |
drone-shot/example_images/06_poster.jpg filter=lfs diff=lfs merge=lfs -text
|
| 126 |
drone-shot/example_images/06_video.mp4 filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 124 |
drone-shot/example_images/05_video.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 125 |
drone-shot/example_images/06_poster.jpg filter=lfs diff=lfs merge=lfs -text
|
| 126 |
drone-shot/example_images/06_video.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 127 |
+
minimax-h3-photorealistic-image-generator-lora-workflow/example_images/01_video.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 128 |
+
minimax-h3-photorealistic-image-generator-lora-workflow/example_images/02_image.jpeg filter=lfs diff=lfs merge=lfs -text
|
| 129 |
+
minimax-h3-photorealistic-image-generator-lora-workflow/example_images/03_image.jpeg filter=lfs diff=lfs merge=lfs -text
|
| 130 |
+
minimax-h3-photorealistic-image-generator-lora-workflow/example_images/04_image.jpeg filter=lfs diff=lfs merge=lfs -text
|
| 131 |
+
minimax-h3-photorealistic-image-generator-lora-workflow/example_images/05_image.jpeg filter=lfs diff=lfs merge=lfs -text
|
| 132 |
+
minimax-h3-photorealistic-image-generator-lora-workflow/example_images/06_image.jpeg filter=lfs diff=lfs merge=lfs -text
|
minimax-h3-photorealistic-image-generator-lora-workflow/example_images/01_poster.jpg
ADDED
|
minimax-h3-photorealistic-image-generator-lora-workflow/example_images/01_video.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0d009a0056f8cceb8e6af3ece902b2cc6d072e6334bb5fe5ca0174afd8e0fd38
|
| 3 |
+
size 2030122
|
minimax-h3-photorealistic-image-generator-lora-workflow/example_images/02_image.jpeg
ADDED
|
Git LFS Details
|
minimax-h3-photorealistic-image-generator-lora-workflow/example_images/03_image.jpeg
ADDED
|
Git LFS Details
|
minimax-h3-photorealistic-image-generator-lora-workflow/example_images/04_image.jpeg
ADDED
|
Git LFS Details
|
minimax-h3-photorealistic-image-generator-lora-workflow/example_images/05_image.jpeg
ADDED
|
Git LFS Details
|
minimax-h3-photorealistic-image-generator-lora-workflow/example_images/06_image.jpeg
ADDED
|
Git LFS Details
|
minimax-h3-photorealistic-image-generator-lora-workflow/h3_photoreal_ph0t0r34l_1024-step00003780.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:446f508a699a0dfe53ec97367eb52e802a88643abec6fb1bc63a5a5b93b7747d
|
| 3 |
+
size 596450960
|
minimax-h3-photorealistic-image-generator-lora-workflow/info.txt
ADDED
|
@@ -0,0 +1,253 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
═══════════════════════════════════════════════════════════════════════════════
|
| 2 |
+
MINIMAX H3 PHOTOREALISTIC IMAGE GENERATOR (LORA + WORKFLOW)
|
| 3 |
+
═══════════════════════════════════════════════════════════════════════════════
|
| 4 |
+
|
| 5 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 6 |
+
IDENTITY
|
| 7 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 8 |
+
Base model MiniMax H3
|
| 9 |
+
Generation mode First-and-Last-Frame-to-Video (FL2VA)
|
| 10 |
+
Checkpoint fl2va (drives T2VA / I2VA / L2VA / FL2VA)
|
| 11 |
+
Category Photorealism
|
| 12 |
+
Author JonXL
|
| 13 |
+
Source https://civitai.com/models/2884206
|
| 14 |
+
Provenance CivitAI · model 2884206 · version 3260165
|
| 15 |
+
|
| 16 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 17 |
+
WEIGHTS
|
| 18 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 19 |
+
File h3_photoreal_ph0t0r34l_1024-step00003780.safetensors
|
| 20 |
+
Version v1.0
|
| 21 |
+
Size 568.8 MB
|
| 22 |
+
Format SafeTensor / unknown precision
|
| 23 |
+
SHA256 446f508a699a0dfe53ec97367eb52e802a88643abec6fb1bc63a5a5b93b7747d
|
| 24 |
+
Published 2026-08-23
|
| 25 |
+
|
| 26 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 27 |
+
TRIGGER WORDS
|
| 28 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 29 |
+
› "ph0t0r34l"
|
| 30 |
+
|
| 31 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 32 |
+
RECOMMENDED SETTINGS
|
| 33 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 34 |
+
Taken verbatim from the author's description:
|
| 35 |
+
|
| 36 |
+
Sampler steps 40
|
| 37 |
+
Resolution 3840x2176
|
| 38 |
+
CFG / guidance 1
|
| 39 |
+
Frame count 22
|
| 40 |
+
|
| 41 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 42 |
+
ADOPTION & POPULARITY
|
| 43 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 44 |
+
Downloads 2,024
|
| 45 |
+
Rating 100% positive (95/95)
|
| 46 |
+
Comments 0
|
| 47 |
+
Published 12 days ago
|
| 48 |
+
Download rate 168.1/day since release
|
| 49 |
+
|
| 50 |
+
Adoption rank #21 of 114 in this collection (High)
|
| 51 |
+
Momentum rank #28 of 114 by download rate (Trending · fresh release)
|
| 52 |
+
|
| 53 |
+
Adoption tier is the percentile of total downloads within this collection;
|
| 54 |
+
momentum is the percentile of downloads-per-day since release. Momentum is
|
| 55 |
+
ranked rather than measured against a fixed rate, because this base model
|
| 56 |
+
is itself new and every LoRA looks fast on an absolute scale. Both numbers
|
| 57 |
+
are CivitAI's, read on the scrape date at the foot of this file.
|
| 58 |
+
|
| 59 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 60 |
+
CHAINING / STACKING
|
| 61 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 62 |
+
Stacks with others yes
|
| 63 |
+
Strength when stacked Use a LOWER LoRA strength when running it together with your own character LoRAs. For the "combining both loras" experiment the author only says to try increasing/decreasing strength and running them at different strengths — no specific stacked values given.
|
| 64 |
+
|
| 65 |
+
What the author says (fact-checked against their own text):
|
| 66 |
+
|
| 67 |
+
Author states the LoRA can be run alongside the user's own character
|
| 68 |
+
LoRAs at lower strength: "The lora when run at lower lora strength can
|
| 69 |
+
also help bring out any of your character loras n a more static
|
| 70 |
+
position." He also floats, as an experiment, "combining both loras with
|
| 71 |
+
different strengths" — but he never says which two LoRAs "both loras"
|
| 72 |
+
refers to, so that is left unspecified. Nothing is stated about load or
|
| 73 |
+
chain order.
|
| 74 |
+
|
| 75 |
+
Requires alongside it:
|
| 76 |
+
! Cicaloo 1 Frame custom node plus its special VAE — required when using
|
| 77 |
+
the Cicaloo 1 Frame workflows: "make sure to clome the custom node and
|
| 78 |
+
download the special VAE, no turbo lora required" (node repo:
|
| 79 |
+
https://github.com/cicalooo/ComfyUI-MM-1Frame). No turbo/speed LoRA is
|
| 80 |
+
needed.
|
| 81 |
+
! Kijai's nodes (ComfyUI-KJNodes) and VideoHelperSuite — hedged: "You may
|
| 82 |
+
need to install some frequently used nodes, mostly Kijai's nodes and
|
| 83 |
+
VideoHelperSuite"
|
| 84 |
+
|
| 85 |
+
Do NOT combine with:
|
| 86 |
+
x Reference model + the Cicaloo 1 Frame workflow — the author states
|
| 87 |
+
reference-model support exists in the default workflow only: "With
|
| 88 |
+
support for reference model in the default workflow (not the 1 Frame
|
| 89 |
+
workflow)"
|
| 90 |
+
|
| 91 |
+
General MiniMax H3 rules that apply here (see REFERENCE.txt):
|
| 92 |
+
|
| 93 |
+
• Load the LoRA with LoraLoaderModelOnly between UNETLoader and your
|
| 94 |
+
sampler; stacking means chaining several of those nodes in series.
|
| 95 |
+
• Stacks normally on top of a single acceleration/turbo LoRA; keep only
|
| 96 |
+
one distill LoRA active in the chain.
|
| 97 |
+
• Chain this only onto an fl2va checkpoint (T2VA / I2VA / L2VA / FL2VA).
|
| 98 |
+
Loading it against ref2va can degrade output silently rather than
|
| 99 |
+
erroring.
|
| 100 |
+
|
| 101 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 102 |
+
PROS
|
| 103 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 104 |
+
+ Well established — 2,024 downloads. [source: CivitAI stats]
|
| 105 |
+
+ Positively received — 95 thumbs-up with no down-votes. [source: CivitAI
|
| 106 |
+
stats]
|
| 107 |
+
+ Explicit trigger word(s) — ph0t0r34l — so the effect can be turned on
|
| 108 |
+
and off from the prompt. [source: model metadata]
|
| 109 |
+
+ Author published concrete recommended settings. [source: description]
|
| 110 |
+
|
| 111 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 112 |
+
CONS & CAVEATS
|
| 113 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 114 |
+
(none recorded)
|
| 115 |
+
|
| 116 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 117 |
+
COMPATIBILITY WARNING
|
| 118 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 119 |
+
MiniMax H3 ships two transformers that share tensor key names: the fl2va
|
| 120 |
+
checkpoint (driving T2VA, I2VA, L2VA and FL2VA) and the separate ref2va
|
| 121 |
+
checkpoint (driving Ref2VA). Loading a LoRA against the wrong family can
|
| 122 |
+
silently degrade output instead of raising an error, so confirm the family
|
| 123 |
+
above matches the checkpoint you have loaded. Likewise, LoRAs built for
|
| 124 |
+
full checkpoints often throw adaln_proj shape errors on pruned ones — match
|
| 125 |
+
the build to your checkpoint.
|
| 126 |
+
|
| 127 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 128 |
+
AUTHOR'S DESCRIPTION
|
| 129 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 130 |
+
Minimax H3 Photorealistic Image Generator (LORA + Workflow)
|
| 131 |
+
Consider to download the workflow (from 'optional downloads') for easier
|
| 132 |
+
image extraction.
|
| 133 |
+
A conceptual lora and workflow to create high resolution photorealistic
|
| 134 |
+
images with Minimax H3.
|
| 135 |
+
Remember image generation on WAN? Now you can generate images with H3 too.
|
| 136 |
+
Photorealistic datasets and special training has made it possible to create
|
| 137 |
+
a photorealistic image generation lora for Minimax-H3, with output up to
|
| 138 |
+
3840x2176.
|
| 139 |
+
With two types of available workflows, the photorealism image lora is used
|
| 140 |
+
to either capture the first and last frame out of a 22 frame video
|
| 141 |
+
(Default) or generates 1 frame (Cicaloo Node+VAE).
|
| 142 |
+
Pick either the 3780 or 3990 step variant. Lora 3990 seems to be more
|
| 143 |
+
vibrant / contrasty and sharper detailed.
|
| 144 |
+
Text 2 Image
|
| 145 |
+
This lora has been originally developed with the intent to generate
|
| 146 |
+
photorealistic images from text.
|
| 147 |
+
|
| 148 |
+
Example from Text to Image
|
| 149 |
+
Image 2 Video
|
| 150 |
+
You can also use the "Image2Video" mode to inject an input image into a new
|
| 151 |
+
output. Results may vary, lora strength may play a factor.
|
| 152 |
+
|
| 153 |
+
All images above were made using the "Default workflow".
|
| 154 |
+
Reference Model: With support for reference model in the default workflow
|
| 155 |
+
(not the 1 Frame workflow), you can also use the ph0t0r34l lora with a
|
| 156 |
+
references.
|
| 157 |
+
|
| 158 |
+
You can try either the FL2VA model or REF2FLVA model, both may yield
|
| 159 |
+
different results.
|
| 160 |
+
|
| 161 |
+
Cicaloo 1 Frame node support
|
| 162 |
+
As of 26.08, two additional workflows using "Cicaloo 1 Frame node" have
|
| 163 |
+
been added.
|
| 164 |
+
Speed up times on my 5090 are about 5 times faster for a 3840x2176 image
|
| 165 |
+
generation at 55 seconds vs 5 minutes+. My recommendation is to use the
|
| 166 |
+
"Default workflow" up to 2560x1440 for best quality and the "Cicaloo 1
|
| 167 |
+
Frame workflow" for renders up to 3840x2176 higher resolution.
|
| 168 |
+
Comparison 4K:
|
| 169 |
+
JXL_IMGEN_DEF_V4b_single_lora workflow
|
| 170 |
+
|
| 171 |
+
0.9 lora strength - cfg 1 - 5 minutes 53 seconds
|
| 172 |
+
JXL_IMGEN_1FRAME_T2I_v4 workflow (CICALOO 1 Frame node - based)
|
| 173 |
+
|
| 174 |
+
1.5 lora str / cfg 1.5 - 55 seconds
|
| 175 |
+
Note: Adjusting lora strength and CFG can have an effect on sharpness
|
| 176 |
+
versus stability.
|
| 177 |
+
Prompt: ph0t0r34l, A twenty four year old blonde female looking out a rain-
|
| 178 |
+
streaked train window, blurred landscape passing by, soft interior
|
| 179 |
+
lighting, introspective and melancholic , depth of field
|
| 180 |
+
Pro tip: The 1 frame gen workflow works quite well at 3840x2176. If you can
|
| 181 |
+
afford it, try running it at 40 steps for best results.
|
| 182 |
+
Authors comment
|
| 183 |
+
Feel free to post your safe for work results on this LoRa's page too!
|
| 184 |
+
Please enjoy making content and experimenting with this photorealistic lora
|
| 185 |
+
and workflow!
|
| 186 |
+
Usage:
|
| 187 |
+
- Preferably, use the following trigger word to activate the image
|
| 188 |
+
generation
|
| 189 |
+
- ph0t0r34l
|
| 190 |
+
|
| 191 |
+
- Seems to work very well on resolutions up to 2560x1440 resolution
|
| 192 |
+
(Default workflow) and 4K when using the Cicaloo workflow.
|
| 193 |
+
|
| 194 |
+
- The lora when run at lower lora strength can also help bring out any of
|
| 195 |
+
your character loras n a more static position.
|
| 196 |
+
|
| 197 |
+
- As an experiment, you could try increasing/decreasing lora strength and
|
| 198 |
+
maybe combining both loras with different strengths. 1.0 lora strength is
|
| 199 |
+
standard for "default" ans 1.5 for the Cicaloo workflow.
|
| 200 |
+
|
| 201 |
+
- Let me know your feedback, as I believe this is the only photorealistic
|
| 202 |
+
image generator that exists for H3 so far :)
|
| 203 |
+
|
| 204 |
+
Known issues:
|
| 205 |
+
- The workflow might complain about an image needed to be loaded, just feed
|
| 206 |
+
it an image (ignore that it will have any effect though)
|
| 207 |
+
|
| 208 |
+
- You may need to install some frequently used nodes, mostly Kijai's nodes
|
| 209 |
+
and VideoHelperSuite
|
| 210 |
+
- https://github.com/kijai/ComfyUI-KJNode
|
| 211 |
+
|
| 212 |
+
- When using the Cicaloo workflows, make sure to clome the custom node and
|
| 213 |
+
download the special VAE, no turbo lora required.
|
| 214 |
+
|
| 215 |
+
- https://github.com/cicalooo/ComfyUI-MM-1Frame
|
| 216 |
+
|
| 217 |
+
Comparisons:
|
| 218 |
+
Left is with lora, right is without lora.
|
| 219 |
+
Right click, open image in new tab, notice the difference in sharpness and
|
| 220 |
+
contrast.
|
| 221 |
+
|
| 222 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 223 |
+
EXAMPLE MEDIA
|
| 224 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 225 |
+
6 preview(s) in ./example_images/ — community results for
|
| 226 |
+
this LoRA. Videos keep a matching _poster.jpg still frame.
|
| 227 |
+
|
| 228 |
+
No prompts are recorded below: CivitAI's public API flags these posts as
|
| 229 |
+
having generation metadata but does not return it, so the prompts cannot be
|
| 230 |
+
scraped. Open the model page to read them.
|
| 231 |
+
|
| 232 |
+
01_video.mp4 (video, 1376x802)
|
| 233 |
+
still: 01_poster.jpg
|
| 234 |
+
02_image.jpeg (image, 2560x1440)
|
| 235 |
+
03_image.jpeg (image, 2560x1440)
|
| 236 |
+
04_image.jpeg (image, 2560x1440)
|
| 237 |
+
05_image.jpeg (image, 2560x1440)
|
| 238 |
+
06_image.jpeg (image, 2560x1440)
|
| 239 |
+
|
| 240 |
+
───────────────────────────��───────────────────────────────────────────────────
|
| 241 |
+
HOW TO USE (COMFYUI)
|
| 242 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 243 |
+
1. Copy the .safetensors file into ComfyUI/models/loras/
|
| 244 |
+
2. Load the matching H3 checkpoint with UNETLoader.
|
| 245 |
+
3. Insert LoraLoaderModelOnly between the loader and your sampler.
|
| 246 |
+
4. Set strength as listed above, then add any trigger word to the prompt.
|
| 247 |
+
5. Keep CFG at 1.0 with BasicGuider unless you know you want real CFG.
|
| 248 |
+
|
| 249 |
+
───────────────────────────────────────────────────────────────────────────────
|
| 250 |
+
Scraped by lorakit on 2026-09-04 23:06 UTC.
|
| 251 |
+
Stats and description are the author's; pros/cons are derived from the
|
| 252 |
+
evidence tagged beside each line.
|
| 253 |
+
───────────────────────────────────────────────────────────────────────────────
|
minimax-h3-photorealistic-image-generator-lora-workflow/metadata.json
ADDED
|
@@ -0,0 +1,139 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"name": "Minimax H3 Photorealistic Image Generator (LORA + Workflow)",
|
| 3 |
+
"slug": "minimax-h3-photorealistic-image-generator-lora-workflow",
|
| 4 |
+
"base_model": "MiniMax H3",
|
| 5 |
+
"media_type": "video",
|
| 6 |
+
"model_id": 2884206,
|
| 7 |
+
"version_id": 3260165,
|
| 8 |
+
"version_name": "v1.0",
|
| 9 |
+
"source": "civitai",
|
| 10 |
+
"url": "https://civitai.com/models/2884206",
|
| 11 |
+
"repo_id": null,
|
| 12 |
+
"file_path": null,
|
| 13 |
+
"author": "JonXL",
|
| 14 |
+
"category": {
|
| 15 |
+
"key": "photoreal",
|
| 16 |
+
"label": "Photorealism"
|
| 17 |
+
},
|
| 18 |
+
"modalities": [
|
| 19 |
+
"fl2v"
|
| 20 |
+
],
|
| 21 |
+
"primary_modality": "fl2v",
|
| 22 |
+
"checkpoint_family": "fl2va",
|
| 23 |
+
"trigger_words": [
|
| 24 |
+
"ph0t0r34l"
|
| 25 |
+
],
|
| 26 |
+
"recommended_settings": {
|
| 27 |
+
"Sampler steps": "40",
|
| 28 |
+
"Resolution": "3840x2176",
|
| 29 |
+
"CFG / guidance": "1",
|
| 30 |
+
"Frame count": "22"
|
| 31 |
+
},
|
| 32 |
+
"chaining": {
|
| 33 |
+
"author": {
|
| 34 |
+
"stacks": "yes",
|
| 35 |
+
"notes": "Author states the LoRA can be run alongside the user's own character LoRAs at lower strength: \"The lora when run at lower lora strength can also help bring out any of your character loras n a more static position.\" He also floats, as an experiment, \"combining both loras with different strengths\" — but he never says which two LoRAs \"both loras\" refers to, so that is left unspecified. Nothing is stated about load or chain order.",
|
| 36 |
+
"avoid": [
|
| 37 |
+
"Reference model + the Cicaloo 1 Frame workflow — the author states reference-model support exists in the default workflow only: \"With support for reference model in the default workflow (not the 1 Frame workflow)\""
|
| 38 |
+
],
|
| 39 |
+
"requires": [
|
| 40 |
+
"Cicaloo 1 Frame custom node plus its special VAE — required when using the Cicaloo 1 Frame workflows: \"make sure to clome the custom node and download the special VAE, no turbo lora required\" (node repo: https://github.com/cicalooo/ComfyUI-MM-1Frame). No turbo/speed LoRA is needed.",
|
| 41 |
+
"Kijai's nodes (ComfyUI-KJNodes) and VideoHelperSuite — hedged: \"You may need to install some frequently used nodes, mostly Kijai's nodes and VideoHelperSuite\""
|
| 42 |
+
],
|
| 43 |
+
"order": "",
|
| 44 |
+
"strength": "Use a LOWER LoRA strength when running it together with your own character LoRAs. For the \"combining both loras\" experiment the author only says to try increasing/decreasing strength and running them at different strengths — no specific stacked values given.",
|
| 45 |
+
"stated": true
|
| 46 |
+
},
|
| 47 |
+
"general_rules": [
|
| 48 |
+
"Load the LoRA with LoraLoaderModelOnly between UNETLoader and your sampler; stacking means chaining several of those nodes in series.",
|
| 49 |
+
"Stacks normally on top of a single acceleration/turbo LoRA; keep only one distill LoRA active in the chain.",
|
| 50 |
+
"Chain this only onto an fl2va checkpoint (T2VA / I2VA / L2VA / FL2VA). Loading it against ref2va can degrade output silently rather than erroring."
|
| 51 |
+
]
|
| 52 |
+
},
|
| 53 |
+
"pros": [
|
| 54 |
+
"Well established — 2,024 downloads. [source: CivitAI stats]",
|
| 55 |
+
"Positively received — 95 thumbs-up with no down-votes. [source: CivitAI stats]",
|
| 56 |
+
"Explicit trigger word(s) — ph0t0r34l — so the effect can be turned on and off from the prompt. [source: model metadata]",
|
| 57 |
+
"Author published concrete recommended settings. [source: description]"
|
| 58 |
+
],
|
| 59 |
+
"cons": [],
|
| 60 |
+
"file": {
|
| 61 |
+
"name": "h3_photoreal_ph0t0r34l_1024-step00003780.safetensors",
|
| 62 |
+
"size_bytes": 596450960,
|
| 63 |
+
"size_mb": 568.82,
|
| 64 |
+
"sha256": "446f508a699a0dfe53ec97367eb52e802a88643abec6fb1bc63a5a5b93b7747d"
|
| 65 |
+
},
|
| 66 |
+
"stats": {
|
| 67 |
+
"downloads": 2024,
|
| 68 |
+
"thumbs_up": 95,
|
| 69 |
+
"thumbs_down": 0
|
| 70 |
+
},
|
| 71 |
+
"popularity": {
|
| 72 |
+
"downloads": 2024,
|
| 73 |
+
"thumbs_up": 95,
|
| 74 |
+
"thumbs_down": 0,
|
| 75 |
+
"comments": 0,
|
| 76 |
+
"approval": "100% positive (95/95)",
|
| 77 |
+
"age_days": 12.0,
|
| 78 |
+
"downloads_per_day": 168.08,
|
| 79 |
+
"rank_by_downloads": 21,
|
| 80 |
+
"rank_by_momentum": 28,
|
| 81 |
+
"of_total": 114,
|
| 82 |
+
"adoption_tier": "High",
|
| 83 |
+
"momentum": "Trending · fresh release"
|
| 84 |
+
},
|
| 85 |
+
"selection_score": 3564.0,
|
| 86 |
+
"selection_reasons": [
|
| 87 |
+
"2,024 downloads",
|
| 88 |
+
"95 thumbs-up",
|
| 89 |
+
"documented triggers + usage notes",
|
| 90 |
+
"top-ranked for its category"
|
| 91 |
+
],
|
| 92 |
+
"example_media": [
|
| 93 |
+
{
|
| 94 |
+
"file": "01_video.mp4",
|
| 95 |
+
"type": "video",
|
| 96 |
+
"resolution": "1376x802",
|
| 97 |
+
"prompt": "",
|
| 98 |
+
"poster": "01_poster.jpg"
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"file": "02_image.jpeg",
|
| 102 |
+
"type": "image",
|
| 103 |
+
"resolution": "2560x1440",
|
| 104 |
+
"prompt": ""
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"file": "03_image.jpeg",
|
| 108 |
+
"type": "image",
|
| 109 |
+
"resolution": "2560x1440",
|
| 110 |
+
"prompt": ""
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"file": "04_image.jpeg",
|
| 114 |
+
"type": "image",
|
| 115 |
+
"resolution": "2560x1440",
|
| 116 |
+
"prompt": ""
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"file": "05_image.jpeg",
|
| 120 |
+
"type": "image",
|
| 121 |
+
"resolution": "2560x1440",
|
| 122 |
+
"prompt": ""
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"file": "06_image.jpeg",
|
| 126 |
+
"type": "image",
|
| 127 |
+
"resolution": "2560x1440",
|
| 128 |
+
"prompt": ""
|
| 129 |
+
}
|
| 130 |
+
],
|
| 131 |
+
"tags": [
|
| 132 |
+
"photorealism",
|
| 133 |
+
"style",
|
| 134 |
+
"image generation",
|
| 135 |
+
"h3 image generator"
|
| 136 |
+
],
|
| 137 |
+
"scraped_at": "2026-09-04T23:06:58+00:00",
|
| 138 |
+
"path": "minimax-h3/minimax-h3-photorealistic-image-generator-lora-workflow"
|
| 139 |
+
}
|