How can you remove fake colors created by AI noise cancellation to make the image look more realistic?

818f5a6048681fa8a8c32bf9db012b3515a3b32d

🧠 ProcessingApproach

1. Create a new layer for AI noise reduction, and place the layer blending mode after AI noise reduction from "normal" to "brightness" above the original layer to eliminate false colors;

2. Creates a brightness mask based on AI-generated denoising layers, controlling the noise reduction range and only denoising gray to dark areas in the image, reducing interference with bright and bright areas.

*For detailed steps, please refer to the images below, in orderRead onThat's it.

🖥️ Supported software:

Topaz Denoised AI & Topaz Photo AI、ACR、LR。

DXO PureRaw's distortion correction effect is inconsistent with ACR and LR, so if the lens has built-in correction files, unless correction is disabled in both DXO PureRaw and ACR, the degree of correction may vary, causing the noise reduction layer and the original layer to not align properly. For lenses with built-in calibration files, after turning off lens correction in ACR, you cannot access the built-in correction files after entering Photoshop.

Of course, if you are using contactless manual lenses or other lenses without built-in calibration files, using DXO PureRaw for noise reduction is also feasible. You just need to disable lens correction in ACR and DXO PureRaw, then enter Photoshop and then set the layer to "Filters- Lens Correction" to call up the lens correction file according to the lens model.

*Platform image compression results in less noticeable visual differences,MaterialsMost of the captured details are also in medium gray. If you're interested, you can try this method in Photoshop by finding materials.

I introduced this method at Panasonic's fifth anniversary full-frame exhibition this year (Guangzhou, Changsha, Beijing). After the event, many friends privately contacted me to request courseware and learn more about this method. Here, I share my thoughts with everyone. The image shows the noise reduction process I commonly use at work, including steps to use noiseware to reduce high-frequency noise. Due to word limits, those who need detailed steps can read the following images in order. If using ACR or LR noise reduction, open both Raw images before and after noise reduction simultaneously, place the denoised Raw on the previous Raw layer, then adjust layer blending mode and create a brightness mask.

901b546df25add6e873c48d58b16b74d895f3660

1ed70a8779600bff3941f8e98c5645fc15e625fe

5a9fd8140b08ffe122e075fa97471585131c48a6

b712a53e4b1f56ed0de729796790d7bd24d97e84

c33e4f2b4bd9fe1197992a716ffc53d6b573cc3b

b1001e9a0786842a3218de1c4afcee8fcd083bd3

5e825834f1e2a7a3d060d33c48a5e54b325ff4c7

1d197dd3b83a1295663805626e0952d67878e2bf

4681ea68b0fbbe3fc554e3a1043573b51a72479f

d74f1dd378f69b4af4208ce27c6e334b9fd9d584

85a2d08e3ca9f3d711d52ca740b28e2bc3764073

65ac776e922ce159bb89d433b16d9348dca68647

📷 The reasons for this are:

1. In low-light environments when shooting at high ISO, the most significant dispersion noise affecting the image mainly occurs in dark and medium gray areas. Bright areas and highlights have higher signal noise due to higher light intake than in dark areas;

2. Compared to dark areas, the human eye tends to be more sensitive to details in bright and highlights, so AI noise reduction creates artifacts and colors in bright and high areas more noticeably than in dark areas and mid-gray;

3. Noiseware is non-AI noise reductionPluginAlthough the noise reduction intensity is not as strong as AI plugins, the effect is very stable, so this noise reduction layer is set to a certain opacity buffer.

If you're not satisfied with the final result, you can still reversibly adjust the mask and layer opacity. Personally, I prefer to call ACR to control the mask range when selecting a mask.

#摄影 ##摄影后期 ##数码日常 #

Original link:https://www.coolapk.com/…

© Copyright Notice
THE END
If you like it, please show your support
Likes0 Share
Commentary Be the First to Comment

Please log in to comment

    No comments yet