flippinbits
commited on
Commit
•
7f95f0e
1
Parent(s):
6f2101c
updating with v1 details
Browse files- README.md +61 -11
- captions/1.txt +1 -0
- captions/10.txt +1 -0
- captions/11.txt +1 -0
- captions/12.txt +1 -0
- captions/13.txt +1 -0
- captions/14.txt +1 -0
- captions/15.txt +1 -0
- captions/16.txt +1 -0
- captions/17.txt +1 -0
- captions/18.txt +1 -0
- captions/19.txt +1 -0
- captions/2.txt +1 -0
- captions/20.txt +1 -0
- captions/21.txt +1 -0
- captions/22.txt +1 -0
- captions/23.txt +1 -0
- captions/3.txt +1 -0
- captions/4.txt +1 -0
- captions/5.txt +1 -0
- captions/6.txt +1 -0
- captions/7.txt +1 -0
- captions/8.txt +1 -0
- captions/9.txt +1 -0
- config.yaml +11 -10
- images/1.webp +0 -0
- images/2.webp +0 -0
- images/3.webp +0 -0
- images/4.webp +0 -0
- images/5.webp +0 -0
- images/6.webp +0 -0
- lora.safetensors +2 -2
README.md
CHANGED
@@ -3,27 +3,77 @@ license: other
|
|
3 |
license_name: flux-1-dev-non-commercial-license
|
4 |
license_link: https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md
|
5 |
language:
|
6 |
-
- en
|
7 |
tags:
|
8 |
-
- flux
|
9 |
-
- diffusers
|
10 |
-
- lora
|
11 |
-
- replicate
|
12 |
base_model: "black-forest-labs/FLUX.1-dev"
|
13 |
pipeline_tag: text-to-image
|
14 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
15 |
---
|
16 |
|
17 |
-
# Flux
|
|
|
|
|
18 |
|
19 |
-
|
20 |
|
21 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
22 |
|
|
|
23 |
|
24 |
-
##
|
25 |
-
You should use `latentpop style` to trigger the image generation.
|
26 |
|
|
|
27 |
|
28 |
## Use it with the [🧨 diffusers library](https://github.com/huggingface/diffusers)
|
29 |
|
|
|
3 |
license_name: flux-1-dev-non-commercial-license
|
4 |
license_link: https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md
|
5 |
language:
|
6 |
+
- en
|
7 |
tags:
|
8 |
+
- flux
|
9 |
+
- diffusers
|
10 |
+
- lora
|
11 |
+
- replicate
|
12 |
base_model: "black-forest-labs/FLUX.1-dev"
|
13 |
pipeline_tag: text-to-image
|
14 |
+
widget:
|
15 |
+
- text: 'screenprint tshirt design, a happy cat holding a sign that says "I LOVE REPLICATE", LNTP illustration style'
|
16 |
+
output:
|
17 |
+
url: "images/1.webp"
|
18 |
+
|
19 |
+
- text: "a t-shirt, LNTP illustration style"
|
20 |
+
output:
|
21 |
+
url: "images/2.webp"
|
22 |
+
|
23 |
+
- text: "a young girl playing piano, yellow background, LNTP illustration style"
|
24 |
+
output:
|
25 |
+
url: "images/3.webp"
|
26 |
+
|
27 |
+
- text: "a book with the words 'Don't Panic!' written on cover, an homage to the hitchhikers guide to the galaxy, LNTP cartoon style"
|
28 |
+
output:
|
29 |
+
url: "images/4.jpg"
|
30 |
+
|
31 |
+
- text: "a robot, blue background, LNTP illustration style"
|
32 |
+
output:
|
33 |
+
url: "images/5.jpg"
|
34 |
+
|
35 |
+
- text: "girl, orange background, LNTP illustration style"
|
36 |
+
output:
|
37 |
+
url: "images/6.webp"
|
38 |
+
|
39 |
+
instance_prompt: LNTP
|
40 |
---
|
41 |
|
42 |
+
# Flux latentpop
|
43 |
+
|
44 |
+
flux-latentpop features vibrant backgrounds with grungy limited screenprinting color goodness.
|
45 |
|
46 |
+
It does great with t-shirt designs, general illustrations, and character portraits.
|
47 |
|
48 |
+
<Gallery />
|
49 |
+
|
50 |
+
It was trained on Replicate, here: https://replicate.com/ostris/flux-dev-lora-trainer/train
|
51 |
+
|
52 |
+
The training set is comprised of 23 images generated on MidJourney using the `--sref 3102110963` and `--personalize 3xdy3qw` flags. You can find the entire training set here: https://replicate.delivery/pbxt/Lg3C1KUPfrRZZvJFaaSTmQ9qtAyXSonLvLSuTuj4Nop9vcSu/2024-08-24-latentpop.zip
|
53 |
+
|
54 |
+
Below are the training parameters I used, which seem to work fairly well for illustration/cartoony Flux LoRAs:
|
55 |
+
|
56 |
+
```
|
57 |
+
{
|
58 |
+
"steps": 1300,
|
59 |
+
"lora_rank": 24,
|
60 |
+
"optimizer": "adamw8bit",
|
61 |
+
"batch_size": 4,
|
62 |
+
"resolution": "512,768,1024",
|
63 |
+
"autocaption": true,
|
64 |
+
"input_images": "https://replicate.delivery/pbxt/Lg3C1KUPfrRZZvJFaaSTmQ9qtAyXSonLvLSuTuj4Nop9vcSu/2024-08-24-latentpop.zip",
|
65 |
+
"trigger_word": "LNTP",
|
66 |
+
"learning_rate": 0.0002,
|
67 |
+
"autocaption_suffix": "LNTP style",
|
68 |
+
"caption_dropout_rate": 0.05,
|
69 |
+
}
|
70 |
+
```
|
71 |
|
72 |
+
Shoutout to @ciguleva on x who originally shared this sref on x: https://x.com/ciguleva/status/1827398343779098720
|
73 |
|
74 |
+
## Usage
|
|
|
75 |
|
76 |
+
You should use `LNTP` to trigger the image generation. The output images look more stylistically interesting with a `guidance_scale` of ~`2.5`.
|
77 |
|
78 |
## Use it with the [🧨 diffusers library](https://github.com/huggingface/diffusers)
|
79 |
|
captions/1.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A digital illustration of a person wearing a green hoodie and a mask with large horns on their head. The person is standing in front of an orange background. The illustration is in a comic book style.
|
captions/10.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A painting of a green tractor with black wheels driving up a sandy hill, in a futuristic style with a sense of adventure, a sense of exploration, a sense of technology, a sense of the future, a science fiction movie scene rendered in the Unreal Engine.
|
captions/11.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A painting of an astronaut in a white suit with a green helmet, sitting in a chair with a green umbilical cord attached to him. The astronaut is wearing a backpack and has a green hose connected to his suit. The scene is set against an orange background. The astronaut is looking upwards, possibly observing something in the distance. The painting is done in a vibrant and colorful style, with bold brush strokes and a sense of movement.
|
captions/12.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon illustration of two alien characters, one with green skin and the other with white skin, standing next to each other and holding a gift. The characters are wearing black clothing and have a playful, whimsical appearance. The scene is drawn in a cartoon style.
|
captions/13.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon illustration of a girl with black hair, wearing a yellow jumpsuit and black shorts, flying through the air. She has a surprised look on her face and is surrounded by three orange stars. The image is in a futuristic style, with a sense of adventure and excitement.
|
captions/14.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon illustration of a young girl wearing a yellow jacket and headphones, sitting on the ground with her eyes closed. The illustration is in a comic book style, with bold colors and exaggerated features. The girl is wearing a yellow and black outfit, and the headphones are prominently displayed on her head. The scene is set against a bright yellow background, which adds to the vibrant and playful nature of the illustration. LNTP style.
|
captions/15.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A painting of a young girl with black hair, wearing a yellow jumpsuit, and holding a red ball. She is sitting on the ground with her mouth open, and there are two more balls in the scene, one on the left side and one on the right. The style of the image is a comic book style.
|
captions/16.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon illustration of a girl wearing an orange hat and orange shirt, holding an orange book. She is surrounded by green plants and a green lizard. The scene is drawn in a black and orange style.
|
captions/17.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A painting of a young girl wearing a green hat and jacket, standing in front of two green dogs. The girl is wearing a backpack and has a surprised look on her face. The dogs are positioned to the left and right of the girl, with one dog closer to the left side and the other dog closer to the right side. The scene is set in a dark background, with the girl and dogs being the main focus of the image. The style of the image is a digital illustration, with a playful and whimsical tone.
|
captions/18.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon illustration of a woman with a green headband, wearing a yellow shirt and black pants. She is looking angry and has a frowning expression. The image is drawn in a comic book style.
|
captions/19.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon illustration of a girl with black hair, wearing an orange hoodie, sitting on the ground and looking up at a planet. The girl is wearing black boots and has a surprised look on her face. The scene is set against a yellow background, with stars scattered throughout the image. The illustration is in a black and white style, giving it a classic and timeless feel.
|
captions/2.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon illustration of a person wearing a black hoodie, a yellow helmet, and a yellow mask. The person is standing in front of an orange background. The illustration has a futuristic style.
|
captions/20.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon illustration of a woman wearing a green hat and apron, holding a bowl and a spoon. The woman has a frowning expression on her face. The image is drawn in a simple, cartoon style.
|
captions/21.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A digital illustration of a young boy wearing a yellow spacesuit, with a red helmet and red boots. He is floating in the air, looking up at a planet. The boy is wearing a patch on his suit, and the scene is set against a black background with stars. The illustration has a playful and whimsical style.
|
captions/22.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon illustration, drawn in a comic book style, features a young boy wearing a spacesuit and helmet, sitting in a space shuttle. The boy has his mouth open, possibly in surprise or excitement. The space shuttle is the main focus of the image, with the boy sitting inside it, and the illustration is rendered in a vibrant orange color. The boy's outfit and the space shuttle are the main subjects of the image, and the overall composition is dynamic and engaging. LNTP style.
|
captions/23.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon illustration of a man with a beard, wearing glasses, and a black shirt. The man is looking upwards with a serious expression on his face. The illustration is in a bright yellow background, giving it a vibrant and lively appearance. The man's beard is described as a goatee, and he is wearing a necklace. The style of the illustration is reminiscent of a comic book or graphic novel.
|
captions/3.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A painting of a person wearing a mask and a hooded jacket, with a yellow lens flare in the background. The person is standing in front of an orange background, and the painting is done in a comic book style. The person is wearing a black jacket and a black mask, with a yellow lens flare in the background. The painting is done in a comic book style, with bold colors and exaggerated features. LNTP style.
|
captions/4.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon illustration of a man wearing a black hoodie, a yellow headband, and green glasses. The man is wearing a black beard and has a frowning expression on his face. The image is drawn in a comic book style.
|
captions/5.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon drawing of a cat wearing a black hat and green skin, with a yellow background. The cat is wearing a black jacket and has a green hat. The cat is looking to the side and appears to be angry. The drawing is in a comic book style.
|
captions/6.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A digital illustration of a black cat wearing a black hat and a black suit. The cat is looking to the side with a green eye and a red nose. The cat is wearing a tie and has a mustache. The background is orange. LNTP style.
|
captions/7.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon illustration of a black cat wearing a wizard's hat, green glasses, and a black coat. The cat is looking to the left, with its whiskers out. The cat is standing in front of a yellow background. The illustration is in a comic book style.
|
captions/8.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon illustration of a cat wearing a black hat with a green background. The cat has a green face and is looking upwards. The cat is wearing a black hat with a yellow emblem on it. The illustration is in a comic book style.
|
captions/9.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
A cartoon illustration of a fat orange mouse with yellow eyes, sitting on a green background. The mouse is wearing black pants and a black shirt. The image is drawn in a comic book style.
|
config.yaml
CHANGED
@@ -5,28 +5,29 @@ config:
|
|
5 |
- type: custom_sd_trainer
|
6 |
training_folder: output
|
7 |
device: cuda:0
|
8 |
-
trigger_word:
|
9 |
network:
|
10 |
type: lora
|
11 |
-
linear:
|
12 |
-
linear_alpha:
|
13 |
save:
|
14 |
dtype: float16
|
15 |
-
save_every:
|
16 |
max_step_saves_to_keep: 1
|
17 |
datasets:
|
18 |
- folder_path: input_images
|
19 |
caption_ext: txt
|
20 |
caption_dropout_rate: 0.05
|
21 |
shuffle_tokens: false
|
22 |
-
cache_latents_to_disk:
|
|
|
23 |
resolution:
|
24 |
- 512
|
25 |
- 768
|
26 |
- 1024
|
27 |
train:
|
28 |
-
batch_size:
|
29 |
-
steps:
|
30 |
gradient_accumulation_steps: 1
|
31 |
train_unet: true
|
32 |
train_text_encoder: false
|
@@ -45,15 +46,15 @@ config:
|
|
45 |
quantize: true
|
46 |
sample:
|
47 |
sampler: flowmatch
|
48 |
-
sample_every:
|
49 |
width: 1024
|
50 |
height: 1024
|
51 |
prompts: []
|
52 |
neg: ''
|
53 |
seed: 42
|
54 |
walk_seed: true
|
55 |
-
guidance_scale:
|
56 |
-
sample_steps:
|
57 |
meta:
|
58 |
name: flux_train_replicate
|
59 |
version: '1.0'
|
|
|
5 |
- type: custom_sd_trainer
|
6 |
training_folder: output
|
7 |
device: cuda:0
|
8 |
+
trigger_word: LNTP
|
9 |
network:
|
10 |
type: lora
|
11 |
+
linear: 24
|
12 |
+
linear_alpha: 24
|
13 |
save:
|
14 |
dtype: float16
|
15 |
+
save_every: 1301
|
16 |
max_step_saves_to_keep: 1
|
17 |
datasets:
|
18 |
- folder_path: input_images
|
19 |
caption_ext: txt
|
20 |
caption_dropout_rate: 0.05
|
21 |
shuffle_tokens: false
|
22 |
+
cache_latents_to_disk: false
|
23 |
+
cache_latents: true
|
24 |
resolution:
|
25 |
- 512
|
26 |
- 768
|
27 |
- 1024
|
28 |
train:
|
29 |
+
batch_size: 4
|
30 |
+
steps: 1300
|
31 |
gradient_accumulation_steps: 1
|
32 |
train_unet: true
|
33 |
train_text_encoder: false
|
|
|
46 |
quantize: true
|
47 |
sample:
|
48 |
sampler: flowmatch
|
49 |
+
sample_every: 1301
|
50 |
width: 1024
|
51 |
height: 1024
|
52 |
prompts: []
|
53 |
neg: ''
|
54 |
seed: 42
|
55 |
walk_seed: true
|
56 |
+
guidance_scale: 3.5
|
57 |
+
sample_steps: 28
|
58 |
meta:
|
59 |
name: flux_train_replicate
|
60 |
version: '1.0'
|
images/1.webp
ADDED
images/2.webp
ADDED
images/3.webp
ADDED
images/4.webp
ADDED
images/5.webp
ADDED
images/6.webp
ADDED
lora.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:44eb014d7f58b0573578c64cba6d53a5824c69f63d3a3096f13a2c35c9483b60
|
3 |
+
size 257887544
|