Text-to-Image
Diffusers
lora
ideogram4
style
vintage
film
vhs
found-footage
1990s
industrial
retro
ps2
low-poly
game-screenshot
2004
vertex-lighting
interlacing
template:diffusion-lora
Instructions to use jmanhype/Ideogram-4.0-LoRAs with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use jmanhype/Ideogram-4.0-LoRAs with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("ideogram-ai/ideogram-4-fp8", dtype=torch.bfloat16, device_map="cuda") pipe.load_lora_weights("jmanhype/Ideogram-4.0-LoRAs") prompt = "VHS_RALLY_95 - 1995 Hi8 camcorder found footage aesthetic" image = pipe(prompt).images[0] - Inference
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -12,6 +12,13 @@ tags:
|
|
| 12 |
- found-footage
|
| 13 |
- 1990s
|
| 14 |
- industrial
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
- template:diffusion-lora
|
| 16 |
widget:
|
| 17 |
- output:
|
|
@@ -32,6 +39,15 @@ widget:
|
|
| 32 |
- output:
|
| 33 |
url: atomic_txt2img_small.png
|
| 34 |
text: 'Vintage nuclear test site - txt2img generation'
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
base_model: ideogram-ai/ideogram-4-fp8
|
| 36 |
instance_prompt: null
|
| 37 |
---
|
|
@@ -102,6 +118,8 @@ Wide shot of a rainy pit stop area, makeshift garage with tools scattered on con
|
|
| 102 |
|
| 103 |
### Ektachrome Style v1 (vintage 1960s/70s Kodak Ektachrome film aesthetic)
|
| 104 |
|
|
|
|
|
|
|
| 105 |
Add the trigger word `ektachrome` in the art_style section of the style description, to be used in combination with appropriate aesthetics and mediums. All samples are generated with the following style_description:
|
| 106 |
|
| 107 |
```json
|
|
@@ -134,9 +152,62 @@ Other examples:
|
|
| 134 |
"art_style":"ektachrome, military test site documentation aesthetic"
|
| 135 |
```
|
| 136 |
|
| 137 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 138 |
|
| 139 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 140 |
|
| 141 |
#### Disclaimer and Important notes
|
| 142 |
The LoRAs in this repository are created for research purposes only and are not intended for commercial use. Some are based on TV series which are protected by copyright. Users utilize these models at their own risk. Users are obligated to comply with copyright laws and applicable regulations. The models have been developed for non-commercial purposes, and it is not my intention to infringe on any copyright. I assume no responsibility for any damages or legal consequences arising from the use of these models. no more no less
|
|
|
|
| 12 |
- found-footage
|
| 13 |
- 1990s
|
| 14 |
- industrial
|
| 15 |
+
- retro
|
| 16 |
+
- ps2
|
| 17 |
+
- low-poly
|
| 18 |
+
- game-screenshot
|
| 19 |
+
- "2004"
|
| 20 |
+
- vertex-lighting
|
| 21 |
+
- interlacing
|
| 22 |
- template:diffusion-lora
|
| 23 |
widget:
|
| 24 |
- output:
|
|
|
|
| 39 |
- output:
|
| 40 |
url: atomic_txt2img_small.png
|
| 41 |
text: 'Vintage nuclear test site - txt2img generation'
|
| 42 |
+
- output:
|
| 43 |
+
url: block_mission_2004/showcase_01_rooftop_los_santos.jpg
|
| 44 |
+
text: 'BLOCK_MISSION_2004 - Rooftop overlooking Los Santos, PS2 framebuffer aesthetic'
|
| 45 |
+
- output:
|
| 46 |
+
url: block_mission_2004/showcase_04_police_chase_night.jpg
|
| 47 |
+
text: 'BLOCK_MISSION_2004 - Police chase downtown at night, 480i interlacing'
|
| 48 |
+
- output:
|
| 49 |
+
url: block_mission_2004/showcase_05_cluckin_bell_interior.jpg
|
| 50 |
+
text: 'BLOCK_MISSION_2004 - Cluckin Bell restaurant interior, vertex lighting'
|
| 51 |
base_model: ideogram-ai/ideogram-4-fp8
|
| 52 |
instance_prompt: null
|
| 53 |
---
|
|
|
|
| 118 |
|
| 119 |
### Ektachrome Style v1 (vintage 1960s/70s Kodak Ektachrome film aesthetic)
|
| 120 |
|
| 121 |
+
**Download:** `ektachrome_style_v1.safetensors` in the [Files & versions tab](/jmanhype/Ideogram-4.0-LoRAs/tree/main)
|
| 122 |
+
|
| 123 |
Add the trigger word `ektachrome` in the art_style section of the style description, to be used in combination with appropriate aesthetics and mediums. All samples are generated with the following style_description:
|
| 124 |
|
| 125 |
```json
|
|
|
|
| 152 |
"art_style":"ektachrome, military test site documentation aesthetic"
|
| 153 |
```
|
| 154 |
|
| 155 |
+
---
|
| 156 |
+
|
| 157 |
+
### BLOCK_MISSION_2004 v1 (2004 PS2 real-time framebuffer aesthetic)
|
| 158 |
+
|
| 159 |
+
**Download:** `block_mission_2004.safetensors` in the [Files & versions tab](/jmanhype/Ideogram-4.0-LoRAs/tree/main/block_mission_2004)
|
| 160 |
+
|
| 161 |
+
A style LoRA that transforms any scene into an authentic 2004-era PS2 game screenshot. Low-poly 3D geometry, compressed textures with visible DCT blocks, vertex lighting with hard banding edges, aggressive jaggies from 480i interlaced output, and the unmistakable warm orange color grading of West Coast crime games.
|
| 162 |
+
|
| 163 |
+
**Training details:**
|
| 164 |
+
- **Base model:** Ideogram 4.0 (FP8)
|
| 165 |
+
- **Trainer:** [ai-toolkit](https://github.com/ostris/ai-toolkit) (flowmatch)
|
| 166 |
+
- **Dataset:** 66 images (17 prompts x 4 images from Google Flow Nano Banana Pro)
|
| 167 |
+
- **Rank:** 32 | **Alpha:** 32
|
| 168 |
+
- **Learning rate:** 1e-4
|
| 169 |
+
- **Steps:** 2,500
|
| 170 |
+
- **Optimizer:** AdamW 8-bit
|
| 171 |
+
- **Format:** Native ai-toolkit safetensors (directly compatible with ComfyUI)
|
| 172 |
|
| 173 |
+
**Recommended settings:**
|
| 174 |
+
- **LoRA strength:** 0.8–0.9 (model and clip)
|
| 175 |
+
- **Sampler:** Euler
|
| 176 |
+
- **Steps:** 20
|
| 177 |
+
- **CFG:** 7.0 base, with CFG Override to 3.0 at 70–100%
|
| 178 |
+
- **Resolution:** 1024×768 (landscape)
|
| 179 |
+
|
| 180 |
+
**Trigger/approach:** This is a **style-only LoRA** — no trigger word needed. Simply load the LoRA and describe your scene. The style will apply the 2004 PS2 framebuffer aesthetic to whatever you generate. Works best with prompts that include these visual descriptors:
|
| 181 |
+
|
| 182 |
+
```
|
| 183 |
+
2004 PS2 real-time cutscene, low-poly, low-res compressed textures,
|
| 184 |
+
vertex lighting with visible banding, baked bloom, jaggies, no anti-aliasing.
|
| 185 |
+
PS2 framebuffer 4:2:2 color. Interlaced 480i combing, MPEG-2 DCT block artifacts.
|
| 186 |
+
Warm orange color grading. No modern photorealism, no remake polish.
|
| 187 |
+
```
|
| 188 |
+
|
| 189 |
+
**Sample prompts:**
|
| 190 |
+
|
| 191 |
+
```
|
| 192 |
+
2004 PS2 real-time cutscene, low-poly West Coast crime game. Protagonist on a rooftop overlooking a sun-drenched city skyline with palm trees, a purple lowrider parked nearby on the concrete. Orange smog haze, vertex lighting banding on building faces, visible jaggies on all edges. Interlaced 480i scanlines, MPEG-2 compression artifacts. Warm color grading. Low-res compressed textures, no anti-aliasing.
|
| 193 |
+
|
| 194 |
+
2004 PS2 real-time cutscene, low-poly. Interior of a fast food restaurant with orange and yellow walls, protagonist at the counter. Fluorescent lighting with vertex banding on floor tiles. Low-poly character models, compressed menu board textures. Jaggies on counter edges. Interlaced 480i, MPEG-2 artifacts. Warm orange tone.
|
| 195 |
+
|
| 196 |
+
2004 PS2 real-time cutscene, low-poly. Nighttime police chase downtown, protagonist fleeing on foot between buildings. Police cruisers with red and blue lights, helicopter spotlight from above. Neon signs on storefronts. Heavy jaggies, vertex lighting on wet pavement, interlacing scanlines. Dark blues and purples with neon highlights.
|
| 197 |
+
```
|
| 198 |
+
|
| 199 |
+
**Showcase:**
|
| 200 |
+
|
| 201 |
+
<img src="https://huggingface.co/jmanhype/Ideogram-4.0-LoRAs/resolve/main/block_mission_2004/showcase_01_rooftop_los_santos.jpg" width="400"/>
|
| 202 |
+
<img src="https://huggingface.co/jmanhype/Ideogram-4.0-LoRAs/resolve/main/block_mission_2004/showcase_04_police_chase_night.jpg" width="400"/>
|
| 203 |
+
|
| 204 |
+
<img src="https://huggingface.co/jmanhype/Ideogram-4.0-LoRAs/resolve/main/block_mission_2004/showcase_02_drive_by_shooting.jpg" width="400"/>
|
| 205 |
+
<img src="https://huggingface.co/jmanhype/Ideogram-4.0-LoRAs/resolve/main/block_mission_2004/showcase_03_grove_street.jpg" width="400"/>
|
| 206 |
+
|
| 207 |
+
<img src="https://huggingface.co/jmanhype/Ideogram-4.0-LoRAs/resolve/main/block_mission_2004/showcase_05_cluckin_bell_interior.jpg" width="400"/>
|
| 208 |
+
<img src="https://huggingface.co/jmanhype/Ideogram-4.0-LoRAs/resolve/main/block_mission_2004/showcase_06_santa_maria_beach.jpg" width="400"/>
|
| 209 |
+
|
| 210 |
+
---
|
| 211 |
|
| 212 |
#### Disclaimer and Important notes
|
| 213 |
The LoRAs in this repository are created for research purposes only and are not intended for commercial use. Some are based on TV series which are protected by copyright. Users utilize these models at their own risk. Users are obligated to comply with copyright laws and applicable regulations. The models have been developed for non-commercial purposes, and it is not my intention to infringe on any copyright. I assume no responsibility for any damages or legal consequences arising from the use of these models. no more no less
|