diff --git a/CHANGELOG.md b/CHANGELOG.md index 2182ca6..56ed408 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -1,3 +1,6 @@ +### v2.3.0 - 2024 Sep.20 +- Refactor + ### v2.2.6 - 2024 Sep.18 - Allow disabling `do_not_save_to_config` to use **Defaults** diff --git a/README.md b/README.md index 6f8e08c..c2a2467 100644 --- a/README.md +++ b/README.md @@ -1,28 +1,99 @@ # SD Webui Vectorscope CC -This is an Extension for the [Automatic1111 Webui](https://github.com/AUTOMATIC1111/stable-diffusion-webui), which performs a kind of **Offset Noise** natively, -allowing you to adjust the brightness, contrast, and color of the generations. +This is an Extension for the [Automatic1111 Webui](https://github.com/AUTOMATIC1111/stable-diffusion-webui), which performs a kind of **Offset Noise** natively during inference, allowing you to adjust the brightness, contrast, and color of the generations. > Also supports both old & new [Forge](https://github.com/lllyasviel/stable-diffusion-webui-forge) -> [Sample Images](#sample-images) +## Example Images + +

+
+Base Image w/o Extension +

+ +
+Infotext + +- **Checkpoint:** [realisticVisionV51](https://civitai.com/models/4201?modelVersionId=130072) +- **Positive Prompt:** `(high quality, best quality), a 4k cinematic photo of a gentleman in suit, street in a city at night, (depth of field, bokeh)` +- **Negative Prompt:** `(low quality, worst quality:1.2), [EasyNegative, EasyNegativeV2]` + +```cpp +Steps: 32, Sampler: DPM++ 2M Karras, CFG scale: 7.5, Seed: 3709157017, Size: 512x512, Denoising strength: 0.5 +Clip skip: 2, Token merging ratio: 0.2, Token merging ratio hr: 0.2, RNG: CPU, NGMS: 4 +Hires upscale: 2, Hires steps: 16, Hires upscaler: 2xNomosUni_esrgan_multijpg +``` + +
+ + + + + + + + + + + + + + + + + + + + + +
VibrantCold"Movie when Mexico"
+
    +
  • Alt: True
    • +
  • Saturation: 1.75
    • +
  • Noise: Ones
    • +
  • Scaling: 1 - Cos
    • +
+ 		+
    +
  • Brightness: -5.0
    • +
  • Contrast: 2.5
    • +
  • Saturation: 0.75
    • +
  • R: -3.0
    • +
  • B: 3.0
    • +
  • Noise: Ones
    • +
  • Scaling: 1 - Sin
    • +
+ 		+
    +
  • Brightness: 2.5
    • +
  • Contrast: -2.5
    • +
  • Saturation: 1.25
    • +
  • R: 1.5
    • +
  • G: 3.0
    • +
  • B: -4.0
    • +
  • Noise: Ones
    • +
  • Scaling: 1 - Sin
    • +
+
## How to Use -After installing this Extension, you will see a new section in both **txt2img** and **img2img** tabs. -Refer to the parameters and sample images below and play around with the values. -> **Note:** Since this modifies the underlying latent noise, the composition may change drastically +> **Note:** Since this Extension modifies the underlying latent tensor, the composition may change drastically depending on the parameters -#### Parameters -- **Enable:** Turn on / off this Extension -- **Alt:** Modify an alternative Tensor instead, causing the effects to be significantly stronger -- **Brightness:** Adjust the overall brightness of the image -- **Contrast:** Adjust the overall contrast of the image -- **Saturation:** Adjust the overall saturation of the image +### Basic Parameters + +- **Enable:** Enable the Extension 💀 +- **Alt:** Cause the Extension effects to be stronger + +
+ Technical Detail + + - This parameter makes the Extension modify the `denoised` Tensor instead of the `x` Tensor + +
+ +- **Brightness**, **Contrast**, **Saturation**: Adjust the overall `brightness` / `contrast` / `saturation` of the image #### Color Channels -- Comes with a Color Wheel for visualization -- You can also click and drag on the Color Wheel to select a color directly - > The color picker isn't 100% accurate for **SDXL**, due to 3 layers of conversions... @@ -51,25 +122,28 @@ Refer to the parameters and sample images below and play around with the values.
-#### Buttons -- **Reset:** Reset all settings to the default values -- **Randomize:** Randomize `Brightness`, `Contrast`, `Saturation`, `R`, `G`, `B` +- The Extension also comes with a Color Wheel for visualization, which you can also click on to pick a color directly + +> The color picker isn't 100% accurate due to multiple layers of conversions... #### Style Presets -- Use the `Dropdown` to select a Style then click **Apply Style** to apply +- To apply a Style, select from the `Dropdown` then click **Apply Style** - To save a Style, enter a name in the `Textbox` then click **Save Style** - To delete a Style, enter the name in the `Textbox` then click **Delete Style** - - *Deleted Style is still in the `styles.json` in case you wish to retrieve it* + - *Style that was deleted is still in the `styles.json` in case you wish to retrieve it* - Click **Refresh Style** to update the `Dropdown` if you edited the `styles.json` manually -#### Advanced Settings -- **Process Hires. fix:** By default, this Extension only functions during the **txt2img** phase, so that **Hires. fix** may "fix" the artifacts introduced during **txt2img**. Enable this to process **Hires. fix** phase too. - - This option does not affect **img2img** -- **Process ADetailer:** Enable to process **[ADetailer](https://github.com/Bing-su/adetailer)** phase too. +### Advanced Parameters + +- **Process Hires. fix:** Enable this option to process during the **Hires. fix** phase too + - By default, this Extension only functions during the regular phase of the `txt2img` mode +- **Process ADetailer:** Enable this option to process during the **[ADetailer](https://github.com/Bing-su/adetailer)** phase too - Will usually cause a square of inconsistent colors +- **Randomize using Seed:** Enable this option to use the current generation `seed` to randomize the basic parameters + - Randomized results will be printed in the console #### Noise Settings -> let `x` denote the Tensor ; let `y` denote the operations +> let **`x`** denote the latent Tensor ; let **`y`** denote the operations - **Straight:** All operations are calculated on the same Tensor - `x += x * y` @@ -77,9 +151,9 @@ Refer to the parameters and sample images below and play around with the values. - `x += x' * y` - **Ones:** All operations are calculated on a Tensor filled with ones - `x += 1 * y` -- **N.Random:** All operations are calculated on a Tensor filled with random values from normal distribution +- **N.Random:** All operations are calculated on a Tensor filled with random values in normal distribution - `x += randn() * y` -- **U.Random:** All operations are calculated on a Tensor filled with random values from uniform distribution +- **U.Random:** All operations are calculated on a Tensor filled with random values in uniform distribution - `x += rand() * y` - **Multi-Res:** All operations are calculated on a Tensor generated with multi-res noise algorithm - `x += multires() * y` @@ -87,153 +161,194 @@ Refer to the parameters and sample images below and play around with the values. - `x += abs(F) * y`

- +

+
+Infotext + +- **Checkpoint:** [realisticVisionV51](https://civitai.com/models/4201?modelVersionId=130072) +- **Positive Prompt:** `(high quality, best quality), a 4k photo of a cute dog running in the snow, mountains, day, (depth of field, bokeh)` +- **Negative Prompt:** `(low quality, worst quality:1.2), [EasyNegative, EasyNegativeV2]` +- **Brightness:** `2.5` +- **Contrast:** `2.5` +- **Alt:** `True` +- **Scaling:** `1 - Cos` + +```cpp +Steps: 24, Sampler: DPM++ 2M Karras, CFG scale: 7.5, Seed: 1257068736, Size: 512x512, Denoising strength: 0.5 +Clip skip: 2, Token merging ratio: 0.2, Token merging ratio hr: 0.2, RNG: CPU, NGMS: 4 +Hires upscale: 1.5, Hires steps: 16, Hires upscaler: 2xNomosUni_esrgan_multijpg +``` + +
+ #### Scaling Settings -By default, this Extension offsets the noise by the same amount each step. -But due to the denoising process, this may produce undesired outcomes such as blurriness at high **Brightness** or noises at low **Brightness**. -Therefore, I added a scaling option to modify the offset amount throughout the process. +By default, this Extension offsets the noise by the same amount every step. But depending on the `Sampler` and `Scheduler` used, and whether `Alt.` was enabled or not, the effects might be too strong during the early or the later phase of the process, which in turn causes artifacts. -> Essentially, the "magnitude" of the default Tensor gets smaller every step, so offsetting by the same amount will have stronger effects at the later steps. - -- **Flat:** Default behavior. Same amount each step. -- **Cos:** Cosine scaling. *(High -> Low)* -- **Sin:** Sine scaling. *(Low -> High)* -- **1 - Cos:** *(Low -> High)* -- **1 - Sin:** *(High -> Low)* +- **Flat:** Default behavior +- **Cos:** Cosine scaling `(High -> Low)` +- **Sin:** Sine scaling `(Low -> High)` +- **1 - Cos:** `(Low -> High)` +- **1 - Sin:** `(High -> Low)`

- +

-## Sample Images -- **Checkpoint:** [Animagine XL V3](https://civitai.com/models/260267) -- **Pos. Prompt:** `[high quality, best quality], 1girl, solo, casual, night, street, city, ` -- **Neg. Prompt:** `lowres, [low quality, worst quality], jpeg` -- `Euler A SGMUniform`; `10 steps`; `2.0 CFG`; **Seed:** `2836968120` -- `Multi-Res Abs.` ; `Cos` +
+Infotext -

-Disabled
- -

+- **Checkpoint:** [realisticVisionV51](https://civitai.com/models/4201?modelVersionId=130072) +- **Positive Prompt:** `(high quality, best quality), a 4k photo of a cute cat standing at a flower field in a park, day, (depth of field, bokeh)` +- **Negative Prompt:** `(low quality, worst quality:1.2), [EasyNegative, EasyNegativeV2]` +- **Alt:** `True` +- **Noise:** `Straight Abs.` -

-Brightness: 2.0 ; Contrast: -0.5 ; Saturation: 1.5
-R: 2.5; G: 1.5; B: -3
- -

+```cpp +Steps: 24, Sampler: DPM++ 2M Karras, CFG scale: 7.5, Seed: 3515074713, Size: 512x512, Denoising strength: 0.5 +Clip skip: 2, Token merging ratio: 0.2, Token merging ratio hr: 0.2, RNG: CPU, NGMS: 4 +Hires upscale: 1.5, Hires steps: 12, Hires upscaler: 2xNomosUni_esrgan_multijpg +``` -

-Brightness: -2.5 ; Contrast: 1 ; Saturation: 0.75
-R: -1.5; G: -1.5; B: 4
- -

+
+ +### Buttons +- **Reset:** Reset all `Basic` and `Advanced` parameters to the default values +- **Randomize:** Randomize the `Brightness`, `Contrast`, `Saturation`, `R`, `G`, `B` parameters ## Roadmap -- [X] Extension Released +- [X] Extension Released! - [X] Add Support for **X/Y/Z Plot** - [X] Implement different **Noise** functions -- [X] Add **Randomize** button -- [X] **Style** Presets +- [X] Implement **Randomize** button +- [X] Implement **Style** Presets - [X] Implement **Color Wheel** & **Color Picker** - [X] Implement better scaling algorithms - [X] Add API Docs -- [X] Append Parameters onto Metadata - - You can enable this in the **Infotext** section of the **Settings** tab -- [X] Add Infotext Support *(by. [catboxanon](https://github.com/catboxanon))* -- [X] ADD **HDR** Script -- [X] Add SDXL Support -- [ ] Add Gradient features - -

-X/Y/Z Plot Support
- -

- -

-Randomize
-
-The value is used as the random seed
You can refer to the console to see the randomized values

+- [X] Append Parameters to Infotext +- [X] Improved Infotext Support *(by. [catboxanon](https://github.com/catboxanon))* +- [X] Add **HDR** Script +- [X] Add Support for **SDXL** +- [ ] Implement Gradient features ## API -You can use this Extension via [API](https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/API) by adding an entry to the `alwayson_scripts` of your payload. -An [example](samples/api_example.json) is provided. -The `args` are sent in the following order in an `array`: +You can use this Extension via [API](https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/API) by adding an entry to the `alwayson_scripts` of your payload. An [example](samples/api_example.json) is provided. The `args` are sent in the following order in an `array`: -- **Enable:** `bool` -- **Alt:** `bool` -- **Brightness:** `float` -- **Contrast:** `float` -- **Saturation:** `float` -- **R:** `float` -- **G:** `float` -- **B:** `float` -- **Process Hires. Fix:** `bool` -- **Noise Settings:** `str` -- **Scaling Settings:** `str` + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
ParameterType
Enablebool
Alt.bool
Brightnessfloat
Contrastfloat
Saturationfloat
Rfloat
Gfloat
Bfloat
Hires. fixbool
ADetailerbool
Randomizebool
Noise Methodstr
Scalingstr
## Known Issues -- Does **not** work certain samplers *(See [Wiki](https://github.com/Haoming02/sd-webui-vectorscope-cc/wiki/Vectorscope-CC-Wiki#effects-with-different-samplers))* -- Has little effect when used with certain **LoRA**s +- In rare occasions, this Extension has little effects when used with certain **LoRA**s +- Works better / worse with certain `Samplers` + ## HDR -

BETA

> [Discussion Thread](https://github.com/Haoming02/sd-webui-vectorscope-cc/issues/16) -- In the **Script** `Dropdown` at the bottom, there is now a new option: **`High Dynamic Range`** +In the **Script** `Dropdown` at the bottom, there is now a new **`High Dynamic Range`** option: + - This script will generate multiple images *("Brackets")* of varying brightness, then merge them into 1 HDR image -- *Do provide feedback in the thread!* -- **Highly Recommended** to use a deterministic sampler and high enough steps. `Euler` *(**not** `Euler a`)* worked well in my experience. +- **(Recommended)** Use a deterministic sampler and high enough steps. `Euler` *(**not** `Euler a`)* works well in my experience #### Settings - **Brackets:** The numer of images to generate - **Gaps:** The brightness difference between each image -- **Automatically Merge:** When enabled, this will merge the images using an `OpenCV` algorithm and save to the `HDR` folder in the `outputs` folder; When disabled, this will return all images to the result section, for when you have a more advanced program such as Photoshop to do the merging. - - All the images are still saved to the `outputs` folder regardless +- **Automatically Merge:** When enabled, this will merge the images using an `OpenCV` algorithm and save to the `HDR` folder in the `outputs` folder + - Disable this if you want to merge them yourself using better external program
Offset Noise TL;DR -The most common *version* of **Offset Noise** you may have heard of is from this [blog post](https://www.crosslabs.org/blog/diffusion-with-offset-noise), -where it was discovered that the noise functions used during **training** were flawed, causing `Stable Diffusion` to always generate images with an average of `0.5` *(**ie.** grey)*. +The most common *version* of **Offset Noise** you may have heard of is from this [blog post](https://www.crosslabs.org/blog/diffusion-with-offset-noise), where it was discovered that the noise functions used during **training** were flawed, causing `Stable Diffusion` to always generate images with an average of `0.5` *(**ie.** grey)*. -> **ie.** Even if you prompt for dark/night or bright/snow, the overall image still looks "grey" +> **ie.** Even if you prompt for dark/night or bright/snow, the average of the image is still "grey" > [Technical Explanations](https://youtu.be/cVxQmbf3q7Q) -However, this Extension instead tries to offset the latent noise during the **inference** phase. -Therefore, you do not need to use models that were specially trained, as this can work on any model. +However, this Extension instead tries to offset the latent noise during the **inference** phase. Therefore, you do not need to use models that were specially trained, as this can work on any model.
How does this work? -After reading through and messing around with the code, -I found out that it is possible to directly modify the Tensors -representing the latent noise used by the Stable Diffusion process. +After reading through and messing around with the code, I found out that it is possible to directly modify the Tensors representing the latent noise used by the Stable Diffusion process. -The dimensions of the Tensors is `(X, 4, H / 8, W / 8)`, which can be thought of like this: - -> **X** batch of noise images, with **4** channels, each with **(W / 8) x (H / 8)** values +The dimensions of the Tensors is `(X, 4, H / 8, W / 8)`, which represents **X** batch of noise images, with **4** channels, each with **(W / 8) x (H / 8)** values > **eg.** Generating a single 512x768 image will create a Tensor of size (1, 4, 96, 64) -Then, I tried to play around with the values of each channel and ended up discovering these relationships. -Essentially, the 4 channels correspond to the **CMYK** color format, -hence why you can control the brightness as well as the colors. +Then, I tried to play around with the values of each channel and ended up discovering these relationships. Essentially, the 4 channels correspond to the **CMYK** color format for `SD1` *(**Y'CbCr** for `SDXL`)*, hence why you can control the brightness as well as the colors.
#### Vectorscope? -The Extension is named this way because the color interactions remind me of the `Vectorscope` found in **Premiere Pro**'s **Lumetri Color**. -Those who are experienced in Color Correction should be rather familiar with this Extension. +The Extension is named this way because the color interactions remind me of the `Vectorscope` found in **Premiere Pro**'s **Lumetri Color**. Those who are experienced in Color Correction should be rather familiar with this Extension.

diff --git a/javascript/vec_cc.js b/javascript/vec_cc.js index c13b867..1ae05c9 100644 --- a/javascript/vec_cc.js +++ b/javascript/vec_cc.js @@ -92,26 +92,17 @@ onUiLoaded(() => { wheel.id = `cc-img-${mode}`; const sliders = [ - [ - document.getElementById(`cc-r-${mode}`).querySelector('input[type=number]'), - document.getElementById(`cc-r-${mode}`).querySelector('input[type=range]') - ], - [ - document.getElementById(`cc-g-${mode}`).querySelector('input[type=number]'), - document.getElementById(`cc-g-${mode}`).querySelector('input[type=range]') - ], - [ - document.getElementById(`cc-b-${mode}`).querySelector('input[type=number]'), - document.getElementById(`cc-b-${mode}`).querySelector('input[type=range]') - ] + document.getElementById(`cc-r-${mode}`).querySelectorAll('input'), + document.getElementById(`cc-g-${mode}`).querySelectorAll('input'), + document.getElementById(`cc-b-${mode}`).querySelectorAll('input'), ]; const temp = document.getElementById(`cc-temp-${mode}`); const dot = temp.querySelector('img'); - dot.id = `cc-dot-${mode}`; dot.style.left = 'calc(50% - 12px)'; dot.style.top = 'calc(50% - 12px)'; + dot.id = `cc-dot-${mode}`; container.appendChild(dot); temp.remove(); diff --git a/lib_cc/__init__.py b/lib_cc/__init__.py index e69de29..cd3f263 100644 --- a/lib_cc/__init__.py +++ b/lib_cc/__init__.py @@ -0,0 +1,4 @@ +""" +Author: Haoming02 +License: MIT +""" diff --git a/lib_cc/callback.py b/lib_cc/callback.py index 31d1147..027596f 100644 --- a/lib_cc/callback.py +++ b/lib_cc/callback.py @@ -38,7 +38,10 @@ class NoiseMethods: @staticmethod @torch.inference_mode() def multires_noise( - latent: torch.Tensor, use_zero: bool, iterations: int = 8, discount: float = 0.4 + latent: torch.Tensor, + use_zero: bool, + iterations: int = 10, + discount: float = 0.8, ): """ Credit: Kohya_SS @@ -46,24 +49,21 @@ class NoiseMethods: """ noise = NoiseMethods.zeros(latent) if use_zero else NoiseMethods.ones(latent) - b, c, w, h = noise.shape - device = latent.device + + b, c, w, h = noise.shape upsampler = torch.nn.Upsample(size=(w, h), mode="bilinear").to(device) - for batch in range(b): - for i in range(iterations): - r = random() * 2 + 2 + for i in range(iterations): + r = random() * 2 + 2 - wn = max(1, int(w / (r**i))) - hn = max(1, int(h / (r**i))) + wn = max(1, int(w / (r**i))) + hn = max(1, int(h / (r**i))) - noise[batch] += ( - upsampler(torch.randn(1, c, hn, wn).to(device)) * discount**i - )[0] + noise += upsampler(torch.randn(b, c, wn, hn).to(device)) * discount**i - if wn == 1 or hn == 1: - break + if wn == 1 or hn == 1: + break return noise / noise.std() @@ -79,6 +79,7 @@ def RGB_2_CbCr(r: float, g: float, b: float) -> tuple[float, float]: original_callback = KDiffusionSampler.callback_state +@torch.no_grad() @torch.inference_mode() @wraps(original_callback) def cc_callback(self, d): @@ -96,7 +97,7 @@ def cc_callback(self, d): mode = str(self.vec_cc["mode"]) method = str(self.vec_cc["method"]) source: torch.Tensor = d[mode] - target: torch.Tensor = None + target = None if "Straight" in method: target = d[mode].detach().clone() @@ -150,7 +151,6 @@ def cc_callback(self, d): source[i][3] *= sat else: - # But why... cb, cr = RGB_2_CbCr(r, g, b) for i in range(batchSize): @@ -170,12 +170,11 @@ def cc_callback(self, d): return original_callback(self, d) -KDiffusionSampler.callback_state = cc_callback - - def restore_callback(): KDiffusionSampler.callback_state = original_callback -on_script_unloaded(restore_callback) -on_ui_settings(settings) +def hook_callbacks(): + KDiffusionSampler.callback_state = cc_callback + on_script_unloaded(restore_callback) + on_ui_settings(settings) diff --git a/lib_cc/colorpicker.py b/lib_cc/colorpicker.py index 4b4eee4..6523a9e 100644 --- a/lib_cc/colorpicker.py +++ b/lib_cc/colorpicker.py @@ -9,16 +9,19 @@ DOT = os.path.join(scripts.basedir(), "scripts", "dot.png") def create_colorpicker(is_img: bool): m: str = "img" if is_img else "txt" - gr.Image( + whl = gr.Image( value=WHEEL, interactive=False, container=False, elem_id=f"cc-colorwheel-{m}", ) - gr.Image( + dot = gr.Image( value=DOT, interactive=False, container=False, elem_id=f"cc-temp-{m}", ) + + whl.do_not_save_to_config = True + dot.do_not_save_to_config = True diff --git a/lib_cc/const.py b/lib_cc/const.py index 0717055..45f1766 100644 --- a/lib_cc/const.py +++ b/lib_cc/const.py @@ -2,6 +2,7 @@ import random class Param: + def __init__(self, minimum: float, maximum: float, default: float): self.minimum = minimum self.maximum = maximum diff --git a/lib_cc/scaling.py b/lib_cc/scaling.py index a934be2..e2a1872 100644 --- a/lib_cc/scaling.py +++ b/lib_cc/scaling.py @@ -13,10 +13,9 @@ def apply_scaling( b: float, ) -> list: - if alg == "Flat": - mod = 1.0 + mod = 1.0 - else: + if alg != "Flat": ratio = float(current_step / total_steps) rad = ratio * pi / 2 diff --git a/lib_cc/style.py b/lib_cc/style.py index e7adc28..d24a07b 100644 --- a/lib_cc/style.py +++ b/lib_cc/style.py @@ -10,6 +10,7 @@ EMPTY_STYLE = {"styles": {}, "deleted": {}} class StyleManager: + def __init__(self): self.STYLE_SHEET: dict = None diff --git a/lib_cc/xyz.py b/lib_cc/xyz.py index 90560af..f67a07b 100644 --- a/lib_cc/xyz.py +++ b/lib_cc/xyz.py @@ -1,7 +1,7 @@ from modules import scripts -def grid_reference(): +def _grid_reference(): for data in scripts.scripts_data: if data.script_class.__module__ in ( "scripts.xyz_grid", @@ -40,7 +40,7 @@ def xyz_support(cache: dict): def choices_scaling(): return ["Flat", "Cos", "Sin", "1 - Cos", "1 - Sin"] - xyz_grid = grid_reference() + xyz_grid = _grid_reference() extra_axis_options = [ xyz_grid.AxisOption( diff --git a/samples/00.jpg b/samples/00.jpg index 8492b57..ee9b464 100644 Binary files a/samples/00.jpg and b/samples/00.jpg differ diff --git a/samples/01.jpg b/samples/01.jpg index 0b003f6..c2ae199 100644 Binary files a/samples/01.jpg and b/samples/01.jpg differ diff --git a/samples/02.jpg b/samples/02.jpg index 79b9bdd..271413a 100644 Binary files a/samples/02.jpg and b/samples/02.jpg differ diff --git a/samples/03.jpg b/samples/03.jpg new file mode 100644 index 0000000..37a8673 Binary files /dev/null and b/samples/03.jpg differ diff --git a/samples/Method.jpg b/samples/Method.jpg deleted file mode 100644 index b3a8b6a..0000000 Binary files a/samples/Method.jpg and /dev/null differ diff --git a/samples/Random.jpg b/samples/Random.jpg deleted file mode 100644 index 0123cfa..0000000 Binary files a/samples/Random.jpg and /dev/null differ diff --git a/samples/Scaling.jpg b/samples/Scaling.jpg deleted file mode 100644 index 4771a38..0000000 Binary files a/samples/Scaling.jpg and /dev/null differ diff --git a/samples/XYZ.jpg b/samples/XYZ.jpg deleted file mode 100644 index e68b904..0000000 Binary files a/samples/XYZ.jpg and /dev/null differ diff --git a/samples/api_example.json b/samples/api_example.json index 8427b79..e3a6cb5 100644 --- a/samples/api_example.json +++ b/samples/api_example.json @@ -1,25 +1,27 @@ { - "prompt": "(masterpiece, best quality), 1girl, solo, night", - "negative_prompt": "(low quality, worst quality:1.2)", - "seed": -1, + "prompt": "a photo of a dog", + "negative_prompt": "(low quality, worst quality)", "sampler_name": "Euler a", "sampler_index": "euler", - "batch_size": 1, "steps": 24, - "cfg_scale": 8.5, + "cfg_scale": 6.0, + "batch_size": 1, + "seed": -1, "width": 512, "height": 512, "alwayson_scripts": { - "Vectorscope CC": { + "vectorscope cc": { "args": [ true, true, - -2.0, + -2.5, 1.5, - 1.25, - 0.0, + 0.85, 0.0, 0.0, + 1.0, + false, + false, false, "Straight Abs.", "Flat" diff --git a/samples/method.jpg b/samples/method.jpg new file mode 100644 index 0000000..c8eefce Binary files /dev/null and b/samples/method.jpg differ diff --git a/samples/scaling.jpg b/samples/scaling.jpg new file mode 100644 index 0000000..c060587 Binary files /dev/null and b/samples/scaling.jpg differ diff --git a/scripts/cc.py b/scripts/cc.py index d769f2b..1be2183 100644 --- a/scripts/cc.py +++ b/scripts/cc.py @@ -2,20 +2,21 @@ from modules.sd_samplers_kdiffusion import KDiffusionSampler from modules import shared, scripts from lib_cc.colorpicker import create_colorpicker +from lib_cc.callback import hook_callbacks from lib_cc.style import StyleManager from lib_cc.xyz import xyz_support -from lib_cc import callback from lib_cc import const from random import seed import gradio as gr -VERSION = "2.2.6" +VERSION = "2.3.0" style_manager = StyleManager() style_manager.load_styles() +hook_callbacks() class VectorscopeCC(scripts.Script): @@ -45,24 +46,24 @@ class VectorscopeCC(scripts.Script): with gr.Row(): bri = gr.Slider( label="Brightness", + value=const.Brightness.default, minimum=const.Brightness.minimum, maximum=const.Brightness.maximum, step=0.05, - value=const.Brightness.default, ) con = gr.Slider( label="Contrast", + value=const.Contrast.default, minimum=const.Contrast.minimum, maximum=const.Contrast.maximum, step=0.05, - value=const.Contrast.default, ) sat = gr.Slider( label="Saturation", + value=const.Saturation.default, minimum=const.Saturation.minimum, maximum=const.Saturation.maximum, step=0.05, - value=const.Saturation.default, ) with gr.Row(): @@ -70,46 +71,37 @@ class VectorscopeCC(scripts.Script): r = gr.Slider( label="R", info="Cyan | Red", + value=const.Color.default, minimum=const.Color.minimum, maximum=const.Color.maximum, step=0.05, - value=const.Color.default, elem_id=f"cc-r-{mode}", ) g = gr.Slider( label="G", info="Magenta | Green", + value=const.Color.default, minimum=const.Color.minimum, maximum=const.Color.maximum, step=0.05, - value=const.Color.default, elem_id=f"cc-g-{mode}", ) b = gr.Slider( label="B", info="Yellow | Blue", + value=const.Color.default, minimum=const.Color.minimum, maximum=const.Color.maximum, step=0.05, - value=const.Color.default, elem_id=f"cc-b-{mode}", ) - r.input( - None, - inputs=[r, g, b], - _js=f"(r, g, b) => {{ VectorscopeCC.updateCursor(r, g, b, {m}); }}", - ) - g.input( - None, - inputs=[r, g, b], - _js=f"(r, g, b) => {{ VectorscopeCC.updateCursor(r, g, b, {m}); }}", - ) - b.input( - None, - inputs=[r, g, b], - _js=f"(r, g, b) => {{ VectorscopeCC.updateCursor(r, g, b, {m}); }}", - ) + for c in (r, g, b): + c.input( + None, + inputs=[r, g, b], + _js=f"(r, g, b) => {{ VectorscopeCC.updateCursor(r, g, b, {m}); }}", + ) create_colorpicker(is_img2img) @@ -125,7 +117,9 @@ class VectorscopeCC(scripts.Script): refresh_btn = gr.Button(value="Refresh Style", scale=2) with gr.Row(elem_classes="style-rows"): - style_name = gr.Textbox(label="Style Name", scale=3) + style_name = gr.Textbox( + label="Style Name", lines=1, max_lines=1, scale=3 + ) save_btn = gr.Button( value="Save Style", elem_id=f"cc-save-{mode}", scale=2 ) @@ -145,12 +139,15 @@ class VectorscopeCC(scripts.Script): with gr.Accordion("Advanced Settings", open=False): with gr.Row(): - doHR = gr.Checkbox(label="Process Hires. fix") + doHR = gr.Checkbox( + label="Process Hires. fix", + visible=(not is_img2img), + ) doAD = gr.Checkbox(label="Process Adetailer") doRN = gr.Checkbox(label="Randomize using Seed") method = gr.Radio( - [ + choices=( "Straight", "Straight Abs.", "Cross", @@ -160,13 +157,13 @@ class VectorscopeCC(scripts.Script): "U.Random", "Multi-Res", "Multi-Res Abs.", - ], + ), label="Noise Settings", value="Straight Abs.", ) scaling = gr.Radio( - ["Flat", "Cos", "Sin", "1 - Cos", "1 - Sin"], + choices=("Flat", "Cos", "Sin", "1 - Cos", "1 - Sin"), label="Scaling Settings", value="Flat", ) @@ -188,7 +185,7 @@ class VectorscopeCC(scripts.Script): apply_btn.click( fn=style_manager.get_style, - inputs=style_choice, + inputs=[style_choice], outputs=[*comps], ).then( None, @@ -199,18 +196,18 @@ class VectorscopeCC(scripts.Script): save_btn.click( fn=lambda *args: gr.update(choices=style_manager.save_style(*args)), inputs=[style_name, *comps], - outputs=style_choice, + outputs=[style_choice], ) delete_btn.click( fn=lambda name: gr.update(choices=style_manager.delete_style(name)), - inputs=style_name, - outputs=style_choice, + inputs=[style_name], + outputs=[style_choice], ) refresh_btn.click( - fn=lambda _: gr.update(choices=style_manager.load_styles()), - outputs=style_choice, + fn=lambda: gr.update(choices=style_manager.load_styles()), + outputs=[style_choice], ) with gr.Row(): @@ -248,7 +245,7 @@ class VectorscopeCC(scripts.Script): outputs=[*comps], show_progress="hidden", ).then( - None, + fn=None, inputs=[r, g, b], _js=f"(r, g, b) => {{ VectorscopeCC.updateCursor(r, g, b, {m}); }}", ) @@ -258,7 +255,7 @@ class VectorscopeCC(scripts.Script): outputs=[bri, con, sat, r, g, b], show_progress="hidden", ).then( - None, + fn=None, inputs=[r, g, b], _js=f"(r, g, b) => {{ VectorscopeCC.updateCursor(r, g, b, {m}); }}", ) @@ -308,52 +305,45 @@ class VectorscopeCC(scripts.Script): seeds: list[int], subseeds: list[int], ): - if "Enable" in self.xyzCache.keys(): - enable = self.xyzCache["Enable"].lower().strip() == "true" + + enable = self.xyzCache.pop("Enable", str(enable)).lower().strip() == "true" if not enable: if len(self.xyzCache) > 0: - if "Enable" not in self.xyzCache.keys(): - print("\n[Vec.CC] x [X/Y/Z Plot] Extension is not Enabled!\n") + print("\n[Vec.CC] x [X/Y/Z Plot] Extension is not Enabled!\n") self.xyzCache.clear() - KDiffusionSampler.vec_cc = {"enable": False} + setattr(KDiffusionSampler, "vec_cc", {"enable": False}) + return p + + method = str(self.xyzCache.pop("Method", method)) + + if method == "Disabled": + setattr(KDiffusionSampler, "vec_cc", {"enable": False}) return p if "Random" in self.xyzCache.keys(): print("[X/Y/Z Plot] x [Vec.CC] Randomize is Enabled.") if len(self.xyzCache) > 1: - print("Some parameters will not apply!") + print("Some parameters will be overridden!") - cc_seed = int(seeds[0]) if doRN else None + cc_seed = int(self.xyzCache.pop("Random")) + else: + cc_seed = int(seeds[0]) if doRN else None - if "Alt" in self.xyzCache.keys(): - latent = self.xyzCache["Alt"].lower().strip() == "true" + latent = self.xyzCache.pop("Alt", str(latent)).lower().strip() == "true" + doHR = self.xyzCache.pop("DoHR", str(doHR)).lower().strip() == "true" + scaling = str(self.xyzCache.pop("Scaling", scaling)) - if "DoHR" in self.xyzCache.keys(): - doHR = self.xyzCache["DoHR"].lower().strip() == "true" + bri = float(self.xyzCache.pop("Brightness", bri)) + con = float(self.xyzCache.pop("Contrast", con)) + sat = float(self.xyzCache.pop("Saturation", sat)) - if "Random" in self.xyzCache.keys(): - cc_seed = int(self.xyzCache["Random"]) + r = float(self.xyzCache.pop("R", r)) + g = float(self.xyzCache.pop("G", g)) + b = float(self.xyzCache.pop("B", b)) - bri = float(self.xyzCache.get("Brightness", bri)) - con = float(self.xyzCache.get("Contrast", con)) - sat = float(self.xyzCache.get("Saturation", sat)) - - r = float(self.xyzCache.get("R", r)) - g = float(self.xyzCache.get("G", g)) - b = float(self.xyzCache.get("B", b)) - - method = str(self.xyzCache.get("Method", method)) - scaling = str(self.xyzCache.get("Scaling", scaling)) - - self.xyzCache.clear() - - if method == "Disabled": - KDiffusionSampler.vec_cc = {"enable": False} - return p - - steps: int = getattr(p, "firstpass_steps", None) or p.steps + assert len(self.xyzCache) == 0 if cc_seed: seed(cc_seed) @@ -366,13 +356,13 @@ class VectorscopeCC(scripts.Script): g = const.Color.rand() b = const.Color.rand() - print(f"\n-> Seed: {cc_seed}") - print(f"Brightness:\t{bri}") - print(f"Contrast:\t{con}") - print(f"Saturation:\t{sat}") - print(f"R:\t\t{r}") - print(f"G:\t\t{g}") - print(f"B:\t\t{b}\n") + print(f"\n[Seed: {cc_seed}]") + print(f"> Brightness: {bri}") + print(f"> Contrast: {con}") + print(f"> Saturation: {sat}") + print(f"> R: {r}") + print(f"> G: {g}") + print(f"> B: {b}\n") if getattr(shared.opts, "cc_metadata", True): p.extra_generation_params.update( @@ -394,6 +384,8 @@ class VectorscopeCC(scripts.Script): } ) + steps: int = getattr(p, "firstpass_steps", None) or p.steps + bri /= steps con /= steps sat = pow(sat, 1.0 / steps) @@ -403,20 +395,32 @@ class VectorscopeCC(scripts.Script): mode: str = "x" if latent else "denoised" - KDiffusionSampler.vec_cc = { - "enable": True, - "mode": mode, - "bri": bri, - "con": con, - "sat": sat, - "r": r, - "g": g, - "b": b, - "method": method, - "doHR": doHR, - "doAD": doAD, - "scaling": scaling, - "step": steps, - } + setattr( + KDiffusionSampler, + "vec_cc", + { + "enable": True, + "mode": mode, + "bri": bri, + "con": con, + "sat": sat, + "r": r, + "g": g, + "b": b, + "method": method, + "doHR": doHR, + "doAD": doAD, + "scaling": scaling, + "step": steps, + }, + ) + + return p + + def before_hr(self, p, enable: bool, *args, **kwargs): + + if enable: + steps: int = getattr(p, "hr_second_pass_steps", None) or p.steps + KDiffusionSampler.vec_cc["step"] = steps return p diff --git a/scripts/cc_hdr.py b/scripts/cc_hdr.py index 0e7dd55..7fc16c5 100644 --- a/scripts/cc_hdr.py +++ b/scripts/cc_hdr.py @@ -1,4 +1,5 @@ from modules.processing import process_images, get_fixed_seed +from modules.shared import state from modules import scripts from copy import copy import gradio as gr @@ -6,39 +7,36 @@ import numpy as np import cv2 as cv -# https://docs.opencv.org/4.8.0/d2/df0/tutorial_py_hdr.html -def merge_HDR(imgs: list, path: str, depth: str, fmt: str, gamma: float) -> np.ndarray: +def _merge_HDR(imgs: list, path: str, depth: str, fmt: str, gamma: float): + """https://docs.opencv.org/4.8.0/d2/df0/tutorial_py_hdr.html""" + import datetime import math import os - output_folder = os.path.join(path, "hdr") - os.makedirs(output_folder, exist_ok=True) + out_dir = os.path.join(os.path.dirname(path), "hdr") + os.makedirs(out_dir, exist_ok=True) + print(f'\nSaving HDR Outputs to "{out_dir}"\n') - imgs_np = [np.asarray(img).astype(np.uint8) for img in imgs] + imgs_np = [np.asarray(img, dtype=np.uint8) for img in imgs] merge = cv.createMergeMertens() hdr = merge.process(imgs_np) - hdr += math.ceil(0 - np.min(hdr) * 1000) / 1000 - # print(f'{np.min(hdr)}, {np.max(hdr)}') + # shift min to 0.0 + hdr += math.ceil(0.0 - np.min(hdr) * 1000) / 1000 + # print(f"({np.min(hdr)}, {np.max(hdr)}") target = 65535 if depth == "16bpc" else 255 - precision = "uint16" if depth == "16bpc" else "uint8" + precision = np.uint16 if depth == "16bpc" else np.uint8 hdr = np.power(hdr, (1 / gamma)) ldr = np.clip(hdr * target, 0, target).astype(precision) rgb = cv.cvtColor(ldr, cv.COLOR_BGR2RGB) - cv.imwrite( - os.path.join( - output_folder, f'{datetime.datetime.now().strftime("%H-%M-%S")}{fmt}' - ), - rgb, - ) - - return ldr + time = datetime.datetime.now().strftime("%H-%M-%S") + cv.imwrite(os.path.join(out_dir, f"{time}{fmt}"), rgb) class VectorHDR(scripts.Script): @@ -50,10 +48,22 @@ class VectorHDR(scripts.Script): return True def ui(self, is_img2img): + with gr.Row(): - count = gr.Slider(label="Brackets", minimum=3, maximum=9, step=2, value=5) + count = gr.Slider( + label="Brackets", + minimum=3, + maximum=9, + step=2, + value=5, + ) + gap = gr.Slider( - label="Gaps", minimum=0.50, maximum=2.50, step=0.25, value=1.25 + label="Gaps", + minimum=0.50, + maximum=2.50, + step=0.25, + value=1.25, ) with gr.Accordion( @@ -61,6 +71,7 @@ class VectorHDR(scripts.Script): elem_id=f'vec-hdr-{"img" if is_img2img else "txt"}', open=False, ): + auto = gr.Checkbox(label="Automatically Merge", value=True) with gr.Row(): @@ -76,7 +87,7 @@ class VectorHDR(scripts.Script): value=1.2, ) - for comp in [count, gap, auto, depth, fmt, gamma]: + for comp in (count, gap, auto, depth, fmt, gamma): comp.do_not_save_to_config = True return [count, gap, auto, depth, fmt, gamma] @@ -84,7 +95,8 @@ class VectorHDR(scripts.Script): def run( self, p, count: int, gap: float, auto: bool, depth: str, fmt: str, gamma: float ): - center = count // 2 + center: int = count // 2 + brackets = brightness_brackets(count, gap) p.seed = get_fixed_seed(p.seed) p.scripts.script("vectorscope cc").xyzCache.update({"Enable": "False"}) @@ -95,9 +107,12 @@ class VectorHDR(scripts.Script): imgs = [None] * count imgs[center] = baseline.images[0] - brackets = brightness_brackets(count, gap) - for it in range(count): + + if state.skipped or state.interrupted or state.stopping_generation: + print("HDR Process Skipped...") + return baseline + if it == center: continue @@ -115,14 +130,13 @@ class VectorHDR(scripts.Script): proc = process_images(pc) imgs[it] = proc.images[0] - if not auto: - baseline.images = imgs - else: - baseline.images = [merge_HDR(imgs, p.outpath_samples, depth, fmt, gamma)] + if auto: + _merge_HDR(imgs, p.outpath_samples, depth, fmt, gamma) + baseline.images = imgs return baseline -def brightness_brackets(count: int, gap: int) -> list[int]: +def brightness_brackets(count: int, gap: float) -> list[float]: half = count // 2 return [gap * (i - half) for i in range(count)]