Skip to content

Commit 0eb8376

Browse files
LoganavterLoganavter
authored
Add Archivist models (#495)
* Add Archiver models * Update author and tags in Archiver Medium model * Change author and update tags in Archiver RGB model Updated author and modified tags in the model JSON. * Change author and update tags in Archiver Rough model Updated author name and modified tags in the JSON model. * Change author and update tags in Archiver Soft model Updated author name and modified tags in the JSON model. * Update 1x-Arhciver-AntiLines.json * Fix filename typo in Archiver-AntiLines.json * Fix author name and update resource size * Fix author casing and update resource size * Fix author name and update model size in JSON * Fix author name and update resource size in JSON * Fix author casing and update resource size
1 parent 38eb945 commit 0eb8376

5 files changed

Lines changed: 225 additions & 0 deletions

File tree

‎data/models/1x-Archiver-AntiLines.json‎

Lines changed: 45 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,45 @@
1+
{
2+
"name": "Archiver AntiLines",
3+
"author": "loganavter",
4+
"license": "MIT",
5+
"tags": [
6+
"denoise",
7+
"restoration",
8+
"dehalo",
9+
"cartoon"
10+
],
11+
"description": "A specialized model from the Archivist suite, designed to remove linear artifacts. It excels at eliminating horizontal lines that other denoisers often mistake for part of the line art.\n\nThis model is optimized for input resolutions between 720p and 1080p. Using it on significantly different resolutions may produce suboptimal results.\n\nAll Archivist models are trained on a custom dataset generated by a physics-based degradation simulator. Recommended Workflow: Use Archivist to fix physical defects, then pass the result through DRUNet (low strength) to stabilize.",
12+
"date": "2025-12-17",
13+
"architecture": "esrgan",
14+
"size": [
15+
"48nf",
16+
"16nb"
17+
],
18+
"scale": 1,
19+
"inputChannels": 3,
20+
"outputChannels": 3,
21+
"resources": [
22+
{
23+
"platform": "pytorch",
24+
"type": "pth",
25+
"size": 1360537,
26+
"sha256": "87140c5284bdecbf15e838d7f30dab5390e02504e47a939180076d6234a82226",
27+
"urls": [
28+
"https://github.com/Loganavter/Archivist-Project-Denoiser/releases/download/v1.0/1x-Archivist_AntiLines.pth"
29+
]
30+
}
31+
],
32+
"trainingIterations": 457000,
33+
"trainingHRSize": 256,
34+
"trainingOTF": true,
35+
"dataset": "Custom dataset (~1800 images) of classic cel animation (1940s-1980s) processed with the Project Degrader physics-based pipeline.",
36+
"datasetSize": 1800,
37+
"images": [
38+
{
39+
"type": "paired",
40+
"caption": "Test case: Removing horizontal film lines",
41+
"LR": "https://imgsli.com/i/d9cdfc24-21c7-4889-926c-c037d453d581.jpg",
42+
"SR": "https://imgsli.com/i/a4f15a21-6196-4e4b-ab88-53b1665b0db4.jpg"
43+
}
44+
]
45+
}

‎data/models/1x-Archiver-Medium.json‎

Lines changed: 45 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,45 @@
1+
{
2+
"name": "Archiver Medium",
3+
"author": "loganavter",
4+
"license": "MIT",
5+
"tags": [
6+
"denoise",
7+
"restoration",
8+
"dehalo",
9+
"cartoon"
10+
],
11+
"description": "A general-purpose model from the Archivist suite, providing a balanced approach to removing common film grain and dirt while preserving the original drawing texture. It is the recommended starting point for most footage.\n\nThis model is optimized for input resolutions between 720p and 1080p. Using it on significantly different resolutions may produce suboptimal results.\n\nAll Archivist models are trained on a custom dataset generated by a physics-based degradation simulator. Recommended Workflow: Use Archivist to fix physical defects, then pass the result through DRUNet (low strength) to stabilize.",
12+
"date": "2025-12-17",
13+
"architecture": "esrgan",
14+
"size": [
15+
"48nf",
16+
"16nb"
17+
],
18+
"scale": 1,
19+
"inputChannels": 3,
20+
"outputChannels": 3,
21+
"resources": [
22+
{
23+
"platform": "pytorch",
24+
"type": "pth",
25+
"size": 1360537,
26+
"sha256": "b94d1fac8873e6bc45cbe8174d0186217a0f2bd4a7a3a103535f6f01972a4b13",
27+
"urls": [
28+
"https://github.com/Loganavter/Archivist-Project-Denoiser/releases/download/v1.0/1x-Archivist_Medium.pth"
29+
]
30+
}
31+
],
32+
"trainingIterations": 478000,
33+
"trainingHRSize": 256,
34+
"trainingOTF": true,
35+
"dataset": "Custom dataset (~1800 images) of classic cel animation (1940s-1980s) processed with the Project Degrader physics-based pipeline.",
36+
"datasetSize": 1800,
37+
"images": [
38+
{
39+
"type": "paired",
40+
"caption": "Test case: Medium noise and grain",
41+
"LR": "https://imgsli.com/i/f9c99bbc-6d87-48fb-9cdf-454115514860.jpg",
42+
"SR": "https://imgsli.com/i/db120500-57d5-4421-a909-c2d31c009ad1.jpg"
43+
}
44+
]
45+
}

‎data/models/1x-Archiver-RGB.json‎

Lines changed: 45 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,45 @@
1+
{
2+
"name": "Archiver RGB",
3+
"author": "loganavter",
4+
"license": "MIT",
5+
"tags": [
6+
"denoise",
7+
"restoration",
8+
"dehalo",
9+
"cartoon"
10+
],
11+
"description": "A specialized model from the Archivist suite, specifically tuned for tackling heavy chromatic (color) noise and severe color channel degradation, often seen in Metrocolor films. Note: its capabilities overlap with the Rough model, but it is better suited for color-based artifacts.\n\nThis model is optimized for input resolutions between 720p and 1080p. Using it on significantly different resolutions may produce suboptimal results.\n\nAll Archivist models are trained on a custom dataset generated by a physics-based degradation simulator. Recommended Workflow: Use Archivist to fix physical defects, then pass the result through DRUNet (low strength) to stabilize.",
12+
"date": "2025-12-17",
13+
"architecture": "esrgan",
14+
"size": [
15+
"48nf",
16+
"16nb"
17+
],
18+
"scale": 1,
19+
"inputChannels": 3,
20+
"outputChannels": 3,
21+
"resources": [
22+
{
23+
"platform": "pytorch",
24+
"type": "pth",
25+
"size": 2721289,
26+
"sha256": "24b2d721cc042d1fb5849625e1e498ea53b6b4d2a7fcddc5d870bcbeb1200b97",
27+
"urls": [
28+
"https://github.com/Loganavter/Archivist-Project-Denoiser/releases/download/v1.0/1x-Archivist_RGB.pth"
29+
]
30+
}
31+
],
32+
"trainingIterations": 193000,
33+
"trainingHRSize": 256,
34+
"trainingOTF": true,
35+
"dataset": "Custom dataset (~1800 images) of classic cel animation (1940s-1980s) processed with the Project Degrader physics-based pipeline.",
36+
"datasetSize": 1800,
37+
"images": [
38+
{
39+
"type": "paired",
40+
"caption": "Test case: Heavy chromatic (color) noise",
41+
"LR": "https://imgsli.com/i/681b8d22-cc4b-41e9-8c4a-b8a829c04533.jpg",
42+
"SR": "https://imgsli.com/i/1ed4a803-cdcb-4e46-87c1-e5776a568389.jpg"
43+
}
44+
]
45+
}

‎data/models/1x-Archiver-Rough.json‎

Lines changed: 45 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,45 @@
1+
{
2+
"name": "Archiver Rough",
3+
"author": "loganavter",
4+
"license": "MIT",
5+
"tags": [
6+
"denoise",
7+
"restoration",
8+
"dehalo",
9+
"cartoon"
10+
],
11+
"description": "An aggressive restoration model from the Archivist suite for severely degraded footage. It attempts to reconstruct heavily damaged or lost details through hallucination. Note: its capabilities overlap with the RGB model, but it focuses more on structural integrity than color noise.\n\nThis model is optimized for input resolutions between 720p and 1080p. Using it on significantly different resolutions may produce suboptimal results.\n\nAll Archivist models are trained on a custom dataset generated by a physics-based degradation simulator. Recommended Workflow: Use Archivist to fix physical defects, then pass the result through DRUNet (low strength) to stabilize.",
12+
"date": "2025-12-17",
13+
"architecture": "esrgan",
14+
"size": [
15+
"48nf",
16+
"16nb"
17+
],
18+
"scale": 1,
19+
"inputChannels": 3,
20+
"outputChannels": 3,
21+
"resources": [
22+
{
23+
"platform": "pytorch",
24+
"type": "pth",
25+
"size": 1360537,
26+
"sha256": "82363da3d4516a8a108e7253ad9056b38e998e22870cc1580ba3e20598d8dbd8",
27+
"urls": [
28+
"https://github.com/Loganavter/Archivist-Project-Denoiser/releases/download/v1.0/1x-Archivist_Rough.pth"
29+
]
30+
}
31+
],
32+
"trainingIterations": 493000,
33+
"trainingHRSize": 256,
34+
"trainingOTF": true,
35+
"dataset": "Custom dataset (~1800 images) of classic cel animation (1940s-1980s) processed with the Project Degrader physics-based pipeline.",
36+
"datasetSize": 1800,
37+
"images": [
38+
{
39+
"type": "paired",
40+
"caption": "Test case: Heavy degradation and noise",
41+
"LR": "https://imgsli.com/i/0f4817f0-69c6-444b-8718-2b39439ef938.jpg",
42+
"SR": "https://imgsli.com/i/339dd3c2-40b5-4da6-9402-6c7688fe0f04.jpg"
43+
}
44+
]
45+
}

‎data/models/1x-Archiver-Soft.json‎

Lines changed: 45 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,45 @@
1+
{
2+
"name": "Archiver Soft",
3+
"author": "loganavter",
4+
"license": "MIT",
5+
"tags": [
6+
"denoise",
7+
"restoration",
8+
"dehalo",
9+
"cartoon"
10+
],
11+
"description": "A light-touch restoration model from the Archivist suite, designed for high-quality sources. It focuses on gentle denoising while preserving the original film grain aesthetic. CAVEAT: In some scenarios, a standard DRUNet might yield subjectively better results. Always compare on your specific footage.\n\nThis model is optimized for input resolutions between 720p and 1080p. Using it on significantly different resolutions may produce suboptimal results.\n\nAll Archivist models are trained on a custom dataset generated by a physics-based degradation simulator. Recommended Workflow: Use Archivist to fix physical defects, then pass the result through DRUNet (low strength) to stabilize.",
12+
"date": "2025-12-17",
13+
"architecture": "esrgan",
14+
"size": [
15+
"48nf",
16+
"16nb"
17+
],
18+
"scale": 1,
19+
"inputChannels": 3,
20+
"outputChannels": 3,
21+
"resources": [
22+
{
23+
"platform": "pytorch",
24+
"type": "pth",
25+
"size": 2721289,
26+
"sha256": "d02616fa398617e2b8ac67cedd5ba265b9ae52372799a75467a9e0d4dd4641f1",
27+
"urls": [
28+
"https://github.com/Loganavter/Archivist-Project-Denoiser/releases/download/v1.0/1x-Archivist_Soft.pth"
29+
]
30+
}
31+
],
32+
"trainingIterations": 453000,
33+
"trainingHRSize": 256,
34+
"trainingOTF": true,
35+
"dataset": "Custom dataset (~1800 images) of classic cel animation (1940s-1980s) processed with the Project Degrader physics-based pipeline.",
36+
"datasetSize": 1800,
37+
"images": [
38+
{
39+
"type": "paired",
40+
"caption": "Test case: Light noise and grain preservation",
41+
"LR": "https://imgsli.com/i/f667ebba-4728-4514-bac6-ccb7e14c03b1.jpg",
42+
"SR": "https://imgsli.com/i/3a42eb2f-f798-4916-b4b7-8eec3ae8d002.jpg"
43+
}
44+
]
45+
}

0 commit comments

Comments
 (0)