Update README.md
Browse files
README.md
CHANGED
@@ -14,20 +14,70 @@ instance_prompt: dreambooth, text to image, bagan
|
|
14 |
license: mit
|
15 |
---
|
16 |
# bagan-text-to-image
|
|
|
|
|
|
|
17 |
|
18 |
-
|
|
|
|
|
|
|
19 |
|
|
|
20 |
|
21 |
-
|
|
|
22 |
|
23 |
-
|
|
|
24 |
|
25 |
-
|
26 |
|
27 |
-
|
28 |
|
|
|
29 |
|
30 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
31 |
|
32 |
|
33 |
-
[Download](/Simbolo-Servicio/bagan-text2image-generator/tree/main) them in the Files & versions tab.
|
|
|
14 |
license: mit
|
15 |
---
|
16 |
# bagan-text-to-image
|
17 |
+
### Text-To-Image (Bagan Ai Generated)
|
18 |
+
### Results
|
19 |
+
All our results are based on fine-tuning https://huggingface.co/runwayml/stable-diffusion-v1-5 model.
|
20 |
|
21 |
+
We display the results using a range of training samples and images from different image categories, such as pagodas and Buddha statues.
|
22 |
+
![Screenshot 2023-10-12 165137](https://github.com/simbolo-ai/Text-to-Image/assets/106800189/fa92aaf2-0346-4563-8d59-17c73440638e)
|
23 |
+
### Ai Generated Bagan Images:
|
24 |
+
![ananda bagan](https://github.com/simbolo-ai/Text-to-Image/assets/106800189/6965b965-73ec-436e-9210-5da550b1fe5b) ![download (2)](https://github.com/simbolo-ai/Text-to-Image/assets/106800189/b1b9af8e-5346-4b52-aa64-2f5839efa9c4)
|
25 |
|
26 |
+
In order to get those images, we trained over 223 General Bagan Images and 137 Ananda Pagodas.
|
27 |
|
28 |
+
### Problem Statement:
|
29 |
+
When we prompted the stable diffusion model to generate an image of Bagan, it produced an image depicting a pagoda from Thailand. Hence, our decision was to fine-tune the current stable diffusion model using a multitude of Bagan photos in order to attain a clearer outcome.
|
30 |
|
31 |
+
### How to use:
|
32 |
+
prompt = "fantasy bagan,hypper detailed , peaceful mood ,The central theme could revolve around a fantastical journey through a magical realm, featuring characters with ethereal and surreal qualities, set against a backdrop of vibrant and enchanting landscapes, The color palette would be a harmonious combination of Jean's bold and surreal hues, by yukisakura sunset."
|
33 |
|
34 |
+
negative_prompt = ""
|
35 |
|
36 |
+
num_samples = 5
|
37 |
|
38 |
+
guidance_scale = 9
|
39 |
|
40 |
+
num_inference_steps = 100
|
41 |
+
|
42 |
+
height = 512
|
43 |
+
|
44 |
+
width = 512
|
45 |
+
|
46 |
+
with autocast("cuda"), torch.inference_mode():
|
47 |
+
images = pipe(
|
48 |
+
prompt,
|
49 |
+
height=height,
|
50 |
+
width=width,
|
51 |
+
negative_prompt=negative_prompt,
|
52 |
+
num_images_per_prompt=num_samples,
|
53 |
+
num_inference_steps=num_inference_steps,
|
54 |
+
guidance_scale=guidance_scale,
|
55 |
+
generator=g_cuda
|
56 |
+
).images
|
57 |
+
|
58 |
+
for img in images:
|
59 |
+
display(img)
|
60 |
+
|
61 |
+
### Data:
|
62 |
+
We used the stable diffusion v1.5 model to train with 223 bagan pictures.
|
63 |
+
|
64 |
+
### Contributors:
|
65 |
+
Main Contributor: [Ye Bhone Lin](https://github.com/Ye-Bhone-Lin)
|
66 |
+
|
67 |
+
Supervisor: Sa Phyo Thu Htet
|
68 |
+
|
69 |
+
Contributors: Thant Htoo San, Min Phone Thit
|
70 |
+
|
71 |
+
### Limitation:
|
72 |
+
We can't generate a photo of a human.
|
73 |
+
|
74 |
+
### References:
|
75 |
+
Wikipedia (2022). Stable Diffusion. Retrieved From: https://en.wikipedia.org/wiki/Stable_Diffusion
|
76 |
+
|
77 |
+
Rombach, R., Blattmann, A., Lorenz, D., Esser, P., & Ommer, B. (2022). High-Resolution Image Synthesis with Latent Diffusion Models. Retrieved From: https://arxiv.org/abs/2112.10752
|
78 |
+
|
79 |
+
Naomi Brown (2022). What is Stable Diffusion and How to Use it. Retrieved From: https://www.fotor.com/blog/what-is-stable-diffusion
|
80 |
+
|
81 |
+
Mishra, O. (June, 9). Stable Diffusion Explained. Medium. https://medium.com/@onkarmishra/stable-diffusion-explained-1f101284484d
|
82 |
|
83 |
|
|