Adaptive sharpen - Image sharpening shader

[bacondither]

It is an image enhancement shader written in HLSL that can be used by MPC-HC or MPDN and other software that can use pixel shaders. Github page (https://github.com/bacondither/Adaptive-sharpen)

Version 2018-04-14
BSD 2-clause license

Download (https://github.com/bacondither/Adaptive-sharpen/archive/2018-04-14.zip)

Changelog:
2018-04-14 - Misc tweaks.

2017-12-30 - Cosmetic changes and minor improvements.

2017-11-01 - Small improvement for anti-ringing.
- Tweaks and minor changes.

2017-04-11 - Use compression instead of clipping for nmax_scale & nmin_scale.
- Tweaked saturation compensation.

2017-01-22 - Replaced the algorithm for saturation compensation.

2016-12-07 - Changed how saturation is compensated for when brightness is altered.
- Misc tweaks.

2016-09-24 - Updated edge-detection weights, slightly better for high-frequency content.
- Minor tweaks.

2016-07-10 - Replaced lerp used for overshoot compression with a weighted power mean, reduces some ringing.
- Corrected minor error in the custom smoothstep function.
- Slightly improved weighting for nmax & nmin.
- Tweaked some parameters.

2016-01-12 - Faster sorting algorithm.

2016-01-07 - Fixed compiler warning X3571 and handle a possible divide by zero.

2015-12-09 - Use a concave sharpening kernel for higher center edge values.
- For summation of pixels in the laplace kernel use a sRGB gamma approximation.

2015-11-17 - Improved low threshold for sharpening.
- Linear light used for summation in the laplace kernel.
- Fixed tanh overflow.

2015-09-23 - Limit low values for the "middle ring" of laplace weights.
- Misc changes.

2015-08-19 - 2-pass algorithm.
- Better overshoot control.
- Reduced saturation loss on pixels made brighter.

2015-06-25 - Unrolls the sorting loop when compiled with ps 4.0+ (much better performance for ps 4.0+).
- Added fix for MPDN (alpha channel is always set to one).
- Style fixes.

2015-05-15 - No real change from the last version, just changed some stupid variable names and comments.

2015-04-05 - First release.



Installation instructions

The shader is tuned and designed for post resize.
Requires gamma light as input and a pixel shader of version 3 or higher.

Settings are stored in the second pass file.
curve_height is the main control for the sharpening strength.

The correct shader order are:
Pass one
Pass two

If the shader passes are added in the wrong order or the alpha channel is truncated, it will return a green screen.


It does what?

The shader tries to sharpen somewhat blurry edges the most, it avoids to sharpen near-flat areas and very sharp edges.
The outer weights of the laplace matrix is variable to mitigate ringing on relative sharp edges and to provide more sharpening on wider and blurrier edges.

The final stage is a soft limiter that confines overshoots based on local near min/max values.
Light overshoots are limited more due to it's worse appearance compared to dark undershoots in most cases.


Old Versions

Depreciated old one pass version:
Download - OLD - 2015-05-25 (https://u.teknik.io/aCpSO7.zip)

[toniash]

It is an image enhancement shader written in HLSL that can be used by MPC-HC or MPDN and other software that can use pixel shaders. It's tuned and designed for post resize and wants gamma light as input.

I will probably rewrite this post soon and add more information!
Meanwhile i hope it works for you :)

BSD 2-clause license

Requires pixel shader version 3.0
Version 2015-04-05

Download (http://www.filedropper.com/adaptivesharpen-2015-04-05)

Very good! :thanks:

What is gamma light?

[Zachs]

Think of Gamma light is the default mode in a render chain. So long as you don't add the ConvertToLinearLight.hlsl file which converts gamma light to linear light, you'll be in gamma light mode.

[KhR0N1K]

awesome work man was waiting for something like this for a long time. it looks amazing. i was testing the FineSharp and LumaSharpen through madVR but i like the way this one looks a lot better.

thanks :D

[foxyshadis]

Testing in MPDN, the image immediately becomes much darker. Is this expected? For some reason, the same thing happens with FineSharp.

Intel 4600, win8.1.

[huhn]

Testing in MPDN, the image immediately becomes much darker. Is this expected? For some reason, the same thing happens with FineSharp.

Intel 4600, win8.1.

works fine for me. even if work used as a pre resize shader.

[KhR0N1K]

Testing in MPDN, the image immediately becomes much darker. Is this expected? For some reason, the same thing happens with FineSharp.

Intel 4600, win8.1.

i noticed this happening when saving screens to compare using MPC-BE. but seemed fine when viewing in the player itself or using PrtScrn instead of Save Image. could be a completely different issue though.

[RainyDog]

Thanks for this bacondither. Tried it last night and liked what I saw.

I'd been using Lumasharpen pre and post resize with reduced clamp as I could never get it to my liking when used in post only. But I found Adaptive Sharpen to be much more effective post only so it might be just what I've been looking for.

Seemed much too aggressive when used as a pre resize sharpener though, but I suppose you mention that it has been tuned for post resize.

[bacondither]

Very good! :thanks:

awesome work man was waiting for something like this for a long time. it looks amazing. i was testing the FineSharp and LumaSharpen through madVR but i like the way this one looks a lot better.

thanks :D
Thanks for this bacondither. Tried it last night and liked what I saw.

I'd been using Lumasharpen pre and post resize with reduced clamp as I could never get it to my liking when used in post only. But I found Adaptive Sharpen to be much more effective post only so it might be just what I've been looking for.

Seemed much too aggressive when used as a pre resize sharpener though, but I suppose you mention that it has been tuned for post resize.

Thanks :)

Testing in MPDN, the image immediately becomes much darker. Is this expected? For some reason, the same thing happens with FineSharp.

Intel 4600, win8.1.

For me it works fine in MPDN.

[Olivier C.]

Hi,

Is there an equivalent of Masktools for GPU ?
Something like mt_merge (merge two layers according to a mask) for example ?

The actual formula is y = ((256 - m) * x1 + m * x2 + 128) / 256 ) where, for a given pixel :
y is the result (merge)
m is the mask value (from 0 to 255)
x1 is the first layer to merge
x2 is the second layer to merge

I am trying to port some avisynth scripts to shaders (GPU) but i am not an expert of .hlsl

Thanks
Olivier

[vivan]

It should work just like that, except everything is done in floating point ([0..1] range).
So
(1.0 - m) * x1 + m * x2 + 0.5
or it's equal
lerp (x1, x2, m) + 0.5

[madshi]

Some caution, though: I think AviSynth scripts usually run in YCbCr, so Y has a range from 16-235. In contrast to that, usually custom pixel shaders are executed in PC levels. So AviSynth 16 should map to floating point 0.0, and AviSynth 235 should map to floating point 1.0. Won't matter for some AviSynth scripts, but it will probably affect some.

[Olivier C.]

It should work just like that, except everything is done in floating point ([0..1] range).
So
(1.0 - m) * x1 + m * x2 + 0.5
or it's equal
lerp (x1, x2, m) + 0.5

Thanks vivan

[Olivier C.]

Some caution, though: I think AviSynth scripts usually run in YCbCr, so Y has a range from 16-235. In contrast to that, usually custom pixel shaders are executed in PC levels. So AviSynth 16 should map to floating point 0.0, and AviSynth 235 should map to floating point 1.0. Won't matter for some AviSynth scripts, but it will probably affect some.

Good point Mathias, thanks !

[madshi]

FYI, I've implemented this in the latest madVR build now, so users can directly play with it, instead of having to setup custom pixel shaders. Thanks for making this available under BSD license!

[KhR0N1K]

this may sound like a dumb question but do i have to have the options "don't use linear light for dithering" and "don't use linear light for smooth motion frame blending" in the madVR options under the "trade quality for performance" tab?

or do these not interfere with the gamma light for the Adaptive Sharpen shader?

also the same question for the "scale in linear light" options for upscaling and downscaling in madVR?

[madshi]

These options are totally unrelated to Adaptive Sharpen. The trade quality options should better be unchecked, unless you need them to be checked for performance reasons. For upscaling linear light is not recommended. For downscaling it's a matter of taste.

[KhR0N1K]

These options are totally unrelated to Adaptive Sharpen. The trade quality options should better be unchecked, unless you need them to be checked for performance reasons. For upscaling linear light is not recommended. For downscaling it's a matter of taste.

ok will do madshi. thanks a lot for the response. i will do what you said and leave them unchecked. and keep the scaling ones off as well.

[foozoor]

Could you make an opengl (glsl) version ?
It would be great for mpv-player users.

[bacondither]

Could you make an opengl (glsl) version ?
It would be great for mpv-player users.
I'm working on improving it now, perhaps in the future i will write a glsl version of it.

[KhR0N1K]

thanks for the experimentals man. i check back here everyday to see if there is improvements :). it is great already however lol.

Edit: the new experimental actually does look better btw. lol. amazing.

[aufkrawall]

Thanks for your efforts, bacondither. AdaptiveSharpen really looks fantastic post-resize with NNEDI3.

[burfadel]

Love this sharpening filter, it works wonders :). I'm using the experimental version. For anyone deciding to try it out, check back occasionally to the first post, as there may be a new experimental version.

[KhR0N1K]

Love this sharpening filter, it works wonders :). I'm using the experimental version. For anyone deciding to try it out, check back occasionally to the first post, as there may be a new experimental version.

exactly i check back every day lol

[burfadel]

I wonder if after a bit more tweaking whether it is found to be suitable to be added to MPDN as a sharpening option?

[aufkrawall]

You can already use it as a render script image processor and with conditionals, you can exactly define when to use it or not.

[burfadel]

I know, just thought it would be easier for people to set up and be aware of :)

[aufkrawall]

Yeah, true.
But dunno if this has priority, as SweetFX with Lumasharpen is already bundled with render scripts, but has to be used with image processor though.

[bacondither]

New version released! :) The previously experimental two pass version is now the stable one.

The experimental version section is removed and replaced with a old versions section.

Now i'm off out in the nice weather!

[burfadel]

Nice! I notice there is a very slight difference in the formula of the second pass between the 28 July version and the latest. Just thought I'd mention that since people may assume that you meant the 28 July version is the same when you said 'The previously experimental two pass version is now the stable one.'

Thanks, it works wonderfully! :)

[KhR0N1K]

New version released! :) The previously experimental two pass version is now the stable one.

The experimental version section is removed and replaced with a old versions section.

Now i'm off out in the nice weather!

awesome man looks and works great thanks for all your hard work.

[lanzorg]

I'm working on improving it now, perhaps in the future i will write a glsl version of it.

That would be really wonderful! :)

[foozoor]

That would be really wonderful! :)

I hope he will hear our vow and that he will realise it. ;)

[burfadel]

Wow, missed there is a new September test build!

Thanks :)!

[burfadel]

Download - EXPERIMENTAL - 2015-09-19 (https://u.teknik.io/k0pyMu.zip)

Thanks, as always :)!

[bacondither]

New version 2015-09-23 released.

Tried optimizing some, it went faster for AMDs old VLIW gpus but slower for newer GCN. So that did not make it to this version.

The bilateral blur from the experimental version did better sometimes and worse other times. But the difference was so minimal and it was slower. So it's not included now.

[aufkrawall]

Thanks for letting us know, interesting to read.

[madshi]

@bacondither, can you double check AdaptiveSharpen with this test image?

http://madshi.net/FineSharpTest3.png

Last time I checked, AdaptiveSharpen makes this image brighter. Which is what most sharpeners do, so it's not surprising. But ideally, that should not happen. Maybe you can work on that? E.g. by using linear light in some parts of your algorithm? Or maybe by running a second pass which "repairs" modified brightness?

Thanks!

[KhR0N1K]

great work and progress bacon looking great and loving the constant upgrades

[aufkrawall]

@bacondither, can you double check AdaptiveSharpen with this test image?

http://madshi.net/FineSharpTest3.png

Last time I checked, AdaptiveSharpen makes this image brighter.
Yes, that's still the case with the current version. Would be great if this could be improved.

[lanzorg]

New version 2015-09-23 released.

Tried optimizing some, it went faster for AMDs old VLIW gpus but slower for newer GCN. So that did not make it to this version.

The bilateral blur from the experimental version did better sometimes and worse other times. But the difference was so minimal and it was slower. So it's not included now.

Could you port this version to glsl?

[ts1]

glsl version here (https://gist.github.com/bjin/e22e96861669e897cf65)
This shader also requires floating point processing apparently.

[vood007]

Here is glsl version of this shader:
Pass 1
...
This shader also requires floating point processing apparently.

Thanks for this. May i request a glsl version of the old one pass version too? Its linked in the first posting.

[ts1]

glsl version of Adaptive Sharpen (https://gist.github.com/igv/8a77e4eb8276753b54bb94c1c50c317e)

[vood007]

vood007

// Copyright (c) 2015, bacondither
// All rights reserved.
...


Thanks again. Works fine under Linux with MPV.

[bacondither]

@bacondither, can you double check AdaptiveSharpen with this test image?

http://madshi.net/FineSharpTest3.png

Last time I checked, AdaptiveSharpen makes this image brighter. Which is what most sharpeners do, so it's not surprising. But ideally, that should not happen. Maybe you can work on that? E.g. by using linear light in some parts of your algorithm? Or maybe by running a second pass which "repairs" modified brightness?

Thanks!

I agree that in some situations local contrast can be increased too much, but sometimes it should get brighter.
For example, if you blur the example image with a gaussian filter or a median filter it gets darker, "unblurring" it then should then make it look brighter.
But the original example image is pretty sharp already, increasing the local contrast/acutance a lot does not look that good in that example.

I have not found a satisfactory solution yet that does not compromise other situations. Sometime in the future i'm going to try testing a third "repair" pass.

I'm pretty sure i'll get some other ideas while in the shower or something... :)

glsl version here (https://gist.github.com/bjin/e22e96861669e897cf65)
This shader also requires floating point processing apparently.

Just had a quick look, i can't understand why is the inputs into tanh(x) is clamped to -+20. That could lead to some screwy results.

[burfadel]

oooh new test version, thanks :)

[ts1]

That is not my port but one of mpv's developers. We made ports in the same day :)

[bacondither]

That is not my port but one of mpv's developers. We made ports in the same day :)

Ah, okay.

I don't have MPV ready for testing now, but did some tests on it a while ago on ubuntu. Could not get some script for switching display frequencies to work as i wanted.

[ts1]

Windows builds (https://mpv.srsfckn.biz/) available also. Just create "portable_config" folder next to mpv.exe with your shaders and mpv.conf with content something like:
vo=opengl-hq:fbo-format=rgba16f:post-shaders="~~/adaptive-sharpen-pass1.glsl,~~/adaptive-sharpen-pass2.glsl"

[madshi]

For example, if you blur the example image with a gaussian filter or a median filter it gets darker
Does it, though, when doing the blurring in linear light? Which is how blurring should ideally be done.

[bacondither]

Does it, though, when doing the blurring in linear light? Which is how blurring should ideally be done.

Yes, forgot that my image editor did blurring in gamma light. Blurred it in linear light and the apparent brightness was then about the same as the original.

Did some tests and saw that doing the summation of the negative laplace vales in "fast" linear light did reduce the brightening effect.
Have not seen any negative effects of doing it that way yet.

Added it to the current experimental version.
Experimental - 2015-10-29 (https://u.teknik.io/qy20mv.zip)

[madshi]

Great, thanks - any progress is welcome! :)

One other thing that I've noticed when comparing FineSharp with AdaptiveSharpen: They have a quite different look. FineSharp somehow makes lines (at least dark lines) appear thinner, while AdaptiveSharpen seems to have a certain tendency to make dark lines fatter. Personally, I like the look of thinner lines. Is that something that could maybe be improved in AdaptiveSharpen?

[aufkrawall]

Added it to the current experimental version.
Experimental - 2015-10-29 (https://u.teknik.io/qy20mv.zip)
Amazing, image brightness pretty much doesn't increase and it still looks as good as usual otherwise. :eek:

Yes, line thickening is noticeable, especially with cartoons. This can look nice with low strength after soft scaling algorithms, but with higher strength it gets a bit too obvious.

[jewshawn2]

I can't seem to get any of the new 2 pass shaders to work with potplayer. However, the early single pass shader works fine.

I use the "Shader Stacker" but no matter what I do I end up with a green screen. But MPC-HC works perfectly fine.

I would much appreciate it if someone else can someone verify?

[toniash]

I can't seem to get any of the new 2 pass shaders to work with potplayer. However, the early single pass shader works fine.

I use the "Shader Stacker" but no matter what I do I end up with a green screen. But MPC-HC works perfectly fine.

I would much appreciate it if someone else can someone verify?
In "trade quality for performance" you must UNTICK:
store custom pixel shader in 16 bit ...
use 10bit chroma ...
use 10bit image ...
run custom pixel shaders in video levels ...

hope it helps

[CruNcher]

Not sure which version is used inside of MPC-BE currently but it does a great job at relatively low shader costs (~10% above the simpler ones) :)

Very impressive results even on rather complex textures but shimmering in high frequency skylines for example still seems to be a big problem at least in the version used inside of MPC-BE.

I use it as a very decent post resize pass solution, which works out for most different compressed MPEG content just great without oversharpen it to much that it feels odd :)

http://i2.sendpic.org/t/3o/3oTpr1c09evN2V1MOXGr3fjtgzi.jpg (http://sendpic.org/view/2/i/55YOJaWChqs3D0KhddFfZEcsXUS.png)http://i2.sendpic.org/t/jl/jlZrMEjPKEbu7nW7xSogt5DcDyH.jpg (http://sendpic.org/view/2/i/j275irS8d6x0ZJRLIaChyt7SFCj.png)http://i1.sendpic.org/t/ph/phnAbVgg8co6HZvj0JiqhKQyem1.jpg (http://sendpic.org/view/1/i/gAB6i4sTRsa8nsVUA03YzGxQY1E.png)

http://i1.sendpic.org/t/95/95kgfqSdffARoOB8FIdoI7qOnhN.jpg (http://sendpic.org/view/1/i/fg1FNBcoQ0RMWjShlUrMc8fbt9B.png)http://i1.sendpic.org/t/g8/g879123LMHEmFsi2oQDCVasEBrn.jpg (http://sendpic.org/view/1/i/2K9hM9OiEJm3t5ute2FXGJY456k.png)http://i2.sendpic.org/t/wO/wO6vOXUZk61HI3XUyU6temyzK3a.jpg (http://sendpic.org/view/2/i/lesT6jOkdWdxIz4k8hmE6G3eUci.png)

http://i1.sendpic.org/t/jY/jYhDRiiQvkE318QNrfnpNXhecuk.jpg (http://sendpic.org/view/1/i/1xnoy2RuMfIldutcu1KPNWu0Lwy.png)http://i1.sendpic.org/t/jx/jxtnRthm3Ob6Fu1k3VP09FCVHf8.jpg (http://sendpic.org/view/1/i/kjPKP0Cvc9vL5vAGCIut9EJmxwr.png)http://i2.sendpic.org/t/aF/aFivjJWFLatrylVIPGpHr0y9Hp9.jpg (http://sendpic.org/view/2/i/8MlbPHAfN3JplvfiK5hfJYlJZmE.png)

http://i2.sendpic.org/t/mw/mwJUh9jvmcRQI02GwidZYKDVBlO.jpg (http://sendpic.org/view/2/i/aRdDz1T4yik8sEUywXZleXvH6Hm.png)http://i2.sendpic.org/t/yq/yqXV82LwbSAWDovpF8MQJkAIS3J.jpg (http://sendpic.org/view/2/i/1BS3XxubBZQlL8vQbkEZjKkXWTq.png)http://i2.sendpic.org/t/u0/u0L7mZYXh8UICgLfgVJ4wythyXC.jpg (http://sendpic.org/view/2/i/9DzsxLo2f2T9FZQLQwXcG9H96aK.png)

[bacondither]

I can't seem to get any of the new 2 pass shaders to work with potplayer. However, the early single pass shader works fine.

I use the "Shader Stacker" but no matter what I do I end up with a green screen. But MPC-HC works perfectly fine.

If you use madvr as renderer you could also try unticking the "lose BTB and WTW if it improves performance"

Not sure which version is used inside of MPC-BE currently
I think it uses the old one-pass 2015-06-25 one.

I could backport some things to the old one pass version and make it better for a very small performance cost.

[bacondither]

The new 2015-11-05 experimental should be a little bit faster then the 2015-10-29 one. ~1% to 2%.

[vood007]

I could backport some things to the old one pass version and make it better for a very small performance cost.

This would be awesome! I still use the older 1-pass version because it allows me to stay at the lowest power-state of my GFX.

[Popwax]

I tried the GLSL port with MPV, but the result is looking weird, could anyone post a sample screenshot (before and after sharpening) of the HLSL shader, so that I can make a comparison?

[bacondither]

Some test images, used the 2015-11-05 shader with default settings.

Blurred test image (http://screenshotcomparison.com/comparison/149622)

AVS HD 709 - Blu-ray & MP4 Calibration, Sharpness & Overscan (http://screenshotcomparison.com/comparison/149624)

Ghost in the shell (http://screenshotcomparison.com/comparison/149626)

[Popwax]

Some test images, used the 2015-11-05 shader with default settings.

Blurred test image (http://screenshotcomparison.com/comparison/149622)

AVS HD 709 - Blu-ray & MP4 Calibration, Sharpness & Overscan (http://screenshotcomparison.com/comparison/149624)

Ghost in the shell (http://screenshotcomparison.com/comparison/149626)

The GLSL port doesn't produce the exact same sharpened image. It widen the edge sigificantly for the first test image, doesn't change anything visiblely for the second one. There are also slight differences for the third test image. Is this due to version different (GLSL port is based on 2015-09 I believe)I? What's the possible cause which can lead to widen edge in the first test image?

[ts1]

Here is mine comparison (this is without scaling):
http://screenshotcomparison.com/comparison/149655

[Popwax]

Here is mine comparison (this is without scaling):
http://screenshotcomparison.com/comparison/149655

Actually what I want to compare are the sharpened images of GLSL and HLSL shader, since I notice distortion of the GLSL shader for noisy images.

bacondither: Could you post more test images for video with heavy film grains?

[ts1]

Yeah, output result is different. MPC-HC vs MPV:
http://screenshotcomparison.com/comparison/149670

[ts1]

This is only with opengl-hq preset (sigmoid-upscaling in particular). With opengl result is identical.

[vood007]

There is some sort of overshot in the glsl (one-pass) version. Upper is HLSL, lower GLSL.

http://thumbnails113.imagebam.com/44542/5d2888445416062.jpg (http://www.imagebam.com/image/5d2888445416062)

[ts1]

vood007
Try updated glsl 1 pass shader.

[vood007]

vood007
Try updated glsl 1 pass shader.

Yep looks fine now, thanks!

[Popwax]

This is only with opengl-hq preset (sigmoid-upscaling in particular). With opengl result is identical.

Oh, that make sense. With "sigmoid-upscaling" option the shader is applied on linear RGB, but the original shader explicitly stated to require gamma light.

EDIT: I can confirm that with "sigmoid-upscaling" disabled, the GLSL shader can produce exactly same sharpened image.

[XMonarchY]

Is there any chance the latest experimental AdaptiveSharpen (11/5/2015) will get an update soon to work properly with DirectX 11? I mainly use AdaptiveSharpen in games through ReShade and right now it says that this latest AdaptiveSharpen does not work as intended in DirectX 11. Does it create some odd artifacts in DirectX 11 or slow down performance too much?

[bacondither]

Is there any chance the latest experimental AdaptiveSharpen (11/5/2015) will get an update soon to work properly with DirectX 11? I mainly use AdaptiveSharpen in games through ReShade and right now it says that this latest AdaptiveSharpen does not work as intended in DirectX 11. Does it create some odd artifacts in DirectX 11 or slow down performance too much?

Hmm.... does it complain about "warning X3081: potentially unintended use of a comma expression in a variable initializer"?

If so, i have already fixed it. It did not result in any errors just the compiler complaining.

[CruNcher]

if i look @ the GPU usage even that small amount 10% hit is huge for games resources.

[JPulowski]

Hmm.... does it complain about "warning X3081: potentially unintended use of a comma expression in a variable initializer"?

If so, i have already fixed it. It did not result in any errors just the compiler complaining.No, the compiler does not give any errors. In DX11, white dot artifacts occur around very thin or very bright surfaces (e.g. stars in the sky, in-game text etc.). Will upload a screenshot when I am available. Not sure what is the problem exactly, ReShade's transcompiler might be the culprit but I don't think that is the case unless there are very tricky shader model specific instructions. Lowering the strength of the shader removes some of the artifacts, but it doesn't make any sense since in DX9 the shader works as it should.

Edit: Here it is, notice the black dots around text and stars:
https://i.imgur.com/9lmTylj.png

[bacondither]

No, the compiler does not give any errors. In DX11, white dot artifacts occur around very thin or very bright surfaces (e.g. stars in the sky, in-game text etc.). Will upload a screenshot when I am available. Not sure what is the problem exactly, ReShade's transcompiler might be the culprit but I don't think that is the case unless there are very tricky shader model specific instructions. Lowering the strength of the shader removes some of the artifacts, but it doesn't make any sense since in DX9 the shader works as it should.
Tested reshade with adaptive sharpen on some dx11 samples.
Found that the problem was with the tanh function that overflowed and resulted in a NaN in dx11 and a 0 in dx9.
It was not really visible in dx9 because it would return 0 when often it should be limited to a very small value near 0.

Modified the AdaptiveSharpen.h so it should work fine now.
Fixed AdaptiveSharpen.h (https://u.teknik.io/r2ASRR.zip)

[bacondither]

New stable version 2015-11-17 and experimental version 2015-11-17-E.

[JPulowski]

Tested reshade with adaptive sharpen on some dx11 samples.
Found that the problem was with the tanh function that overflowed and resulted in a NaN in dx11 and a 0 in dx9.
It was not really visible in dx9 because it would return 0 when often it should be limited to a very small value near 0.

Modified the AdaptiveSharpen.h so it should work fine now.
Fixed AdaptiveSharpen.h (https://u.teknik.io/r2ASRR.zip)Thanks, I have also uploaded the fix to ReShade's repository.

Would you mind telling me which part of the code have you modified? Or did you completely fix it, so there is no need to apply a hotfix after the code has been ported?

[bacondither]

Thanks, I have also uploaded the fix to ReShade's repository.

Would you mind telling me which part of the code have you modified? Or did you completely fix it, so there is no need to apply a hotfix after the code has been ported?

Updated everything to version 2015-11-17. The code that fixed the overflow can be found at line 61 and 243-244 in the new AdaptiveSharpen.h.

Diff for AdaptiveSharpen.h (https://www.diffchecker.com/q9m0ezvb)

[aufkrawall]

I should have looked at the image in unscaled size to judge brightness, sorry.
Don't want to sound like a broken record, but maybe you could still do something regarding brightness?
It still gets noticeably brighter with a strength of 0.5.

off:
http://abload.de/thumb/offeak8l.png (http://abload.de/image.php?img=offeak8l.png)

0.5:
http://abload.de/thumb/0.5n8k5p.png (http://abload.de/image.php?img=0.5n8k5p.png)

[bacondither]

Don't want to sound like a broken record, but maybe you could still do something regarding brightness?
Tuning it to one scenario would make it pretty crappy in other places, i have currently no idea to "solve" it with an algorithm that would work in shaders or be fast.
But the latest version 2015-12-04-E should be somewhat better at fine details in most cases. Uploaded a comparison of 2015-11-17E and 2015-12-04E (http://screenshotcomparison.com/comparison/153261) with all settings at default.

I also experimented some with sharpening that test image in raw therapee with RL deconvolution, that made the image alot more brighter!

[aufkrawall]

This is a very nice progress, thanks a lot!
I think this can be very helpful in games, e.g. when strong sharpen is needed to fight temporal AA blur. May be a bit expensive in realtime, but should be great for screenshots.

[JPulowski]

In games, AdaptiveSharpen requires quite a bit of performance. In my GPU it utilizes 15 - 20% of extra computing power. In movies where the framerate is between 23.976 - 30.0 fps, the performance drop does not make a huge difference. But in recent video games, every bit of performance is important. bacondither, do you have any ideas about possible optimizations? Currently the shader uses an additional RGBA16F texture for storing the alpha value of the first pass. This is a bit too much just for one floating point value. I might use R16F or encode the value as RGBA8 and decode it in the second pass afterwards, depending on which method is faster. Other than that, it is now my favorite shader for sharpening. It really goes well with the video games.

We also have a GitHub repo (https://github.com/crosire/reshade-shaders). You are welcome to modify the code as you see fit. :)

[ts1]

I've ported some things from 2-pass shaders.

Edit: fixed a bug

[foozoor]

I've ported some things from 2-pass shaders.

Edit: fixed a bug

Where can we download it? Can I use it in mpv?
Thanks I love this sharpener but I think 1.0 is too much as default value.

Edit: I found it inside the mpv wiki... :rolleyes:
https://gist.github.com/bjin/152a2f8bdf87d83261ac

[ts1]

foozoor
1-pass version is not so strong, try it, it's on a 3rd page.

[foozoor]

foozoor
1-pass version is not so strong, try it, it's on a 3rd page.
Thank you, really nice. :)

Is there something between lumasharpen and adaptive sharpen for mpv?
Because my gpu can‘t handle adaptive sharpen with some video files.
Could you port finesharp or lsfmod as mpv shader?

[ts1]

Sorry, but no.
Use it as prescale then maybe.

[lanzorg]

foozoor
1-pass version is not so strong, try it, it's on a 3rd page.
Should I turn `linear-scaling` and `sigmoid-upscaling` off like the 2-passes shader?

[ts1]

lanzorg
Yes

[bacondither]

Do you have any ideas about possible optimizations?

Sure, did a pull request on reshade-shaders now. Up to 45% speedup on rendering times for the shader (3.2ms to 2.2ms) at 1080p on my 7790. :p

[aufkrawall]

With a 980, I can note a gain. However, it's not large.
No ReShade shaders: 96fps
AS old: 81fps
AS new: 83fps

[bacondither]

With a 980, I can note a gain. However, it's not large.

At least that's 18% faster rendering times for the shader.
Did you have have "fast_ops = 1" in JPulowski.cfg?

I did my test on a DX11 target application.

[aufkrawall]

Yep, had it at 1 and tested DX11.
With a higher resolution, the results don't get wider:
42 vs 41 fps :)

[ts1]

I've made my port compatible with linear-scaling/sigmoid-upscaling. Also much less distortions now (if any).

[lanzorg]

I've made my port compatible with linear-scaling/sigmoid-upscaling. Also much less distortions now (if any).

Thanks! :D

[darkasylum666]

// Copyright (c) 2015, bacondither
// All rights reserved.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions
// are met:
// 1. Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer
// in this position and unchanged.
// 2. Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
//
// THIS SOFTWARE IS PROVIDED BY THE AUTHORS ``AS IS'' AND ANY EXPRESS OR
// IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
// OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
// IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
// INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
// NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
// THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

// Adaptive sharpen - version 2015-06-25 - (requires ps >= 3.0)
// Tuned for use post resize, EXPECTS FULL RANGE GAMMA LIGHT

#pragma optionNV (unroll all)

#define curve_height 1.0 // Main sharpening strength, POSITIVE VALUE ONLY!
// 0.3 <-> 1.5 is a reasonable range of values

#define video_level_out false // True to preserve BTB & WTW (minor summation error)
// Normally it should be set to false

// Defined values under this row are "optimal" DO NOT CHANGE IF YOU DO NOT KNOW WHAT YOU ARE DOING!

#define curveslope (curve_height*0.7) // Sharpening curve slope, edge region
#define D_overshoot 0.001 // Max dark overshoot before max compression
#define D_comp_ratio 0.250 // Max compression ratio, dark overshoot (1/0.25=4x)
#define L_overshoot 0.003 // Max light overshoot before max compression
#define L_comp_ratio 0.167 // Max compression ratio, light overshoot (1/0.167=6x)
#define max_scale_lim 0.1 // Abs change before max compression (1/10=±10%)

// Saturation loss reduction
#define minim_satloss ( (c[0].rgb*(CtL(c[0].rgb + sharpdiff)/c0_Y) + (c[0].rgb + sharpdiff))/2 )

// Soft limit
#define soft_lim(v,s) ( ((exp(2*min(abs(v), s*16)/s) - 1)/(exp(2*min(abs(v), s*16)/s) + 1))*s )

// Get destination pixel values
#define get(x,y) ( texture(tex, pos + vec2(x*px, y*py)) )
#define sat(input) ( vec3(clamp((input).xyz, 0.0, 1.0) ) )

// Colour to luma
#define CtL(RGB) ( pow(dot(vec3(0.256, 0.651, 0.093), clamp((RGB).rgb, 0.0, 1.0)), 1.2) )

// Center pixel diff
#define mdiff(a,b,c,d,e,f,g) ( abs(luma[g]-luma[a]) + abs(luma[g]-luma[b]) \
+ abs(luma[g]-luma[c]) + abs(luma[g]-luma[d]) \
+ 0.5*(abs(luma[g]-luma[e]) + abs(luma[g]-luma[f])) )

vec4 sample(sampler2D tex, vec2 pos, vec2 tex_size) {
// Pixel "width"
float px = 1.0 / tex_size[0];
float py = 1.0 / tex_size[1];

vec4 orig = texture(tex, pos);

// Get points and saturate out of range values (BTB & WTW)
// [ c22 ]
// [ c24, c9, c23 ]
// [ c21, c1, c2, c3, c18 ]
// [ c19, c10, c4, c0, c5, c11, c16 ]
// [ c20, c6, c7, c8, c17 ]
// [ c15, c12, c14 ]
// [ c13 ]
vec3 c[25] = vec3[](sat( orig), get(-1,-1), get( 0,-1), get( 1,-1), get(-1, 0),
get( 1, 0), get(-1, 1), get( 0, 1), get( 1, 1), get( 0,-2),
get(-2, 0), get( 2, 0), get( 0, 2), get( 0, 3), get( 1, 2),
get(-1, 2), get( 3, 0), get( 2, 1), get( 2,-1), get(-3, 0),
get(-2, 1), get(-2,-1), get( 0,-3), get( 1,-2), get(-1,-2));

// Blur, gauss 3x3
vec3 blur = (2*(c[2]+c[4]+c[5]+c[7]) + (c[1]+c[3]+c[6]+c[8]) + 4*c[0])/16;
float blur_Y = (blur.r*(1.0/3.0) + blur.g*(1.0/3.0) + blur.b*(1.0/3.0));

// Contrast compression, center = 0.5, scaled to 1/3
float c_comp = clamp(0.266666681f + 0.9*pow(2, (-7.4*blur_Y)), 0.0, 1.0);

// Edge detection
// Matrix, relative weights
// [ 1 ]
// [ 4, 4, 4 ]
// [ 1, 4, 4, 4, 1 ]
// [ 4, 4, 4 ]
// [ 1 ]
float edge = length( abs(blur-c[0]) + abs(blur-c[1]) + abs(blur-c[2]) + abs(blur-c[3])
+ abs(blur-c[4]) + abs(blur-c[5]) + abs(blur-c[6]) + abs(blur-c[7]) + abs(blur-c[8])
+ 0.25*(abs(blur-c[9]) + abs(blur-c[10]) + abs(blur-c[11]) + abs(blur-c[12])) );

// Edge detect contrast compression, center = 0.5
edge *= c_comp;

// RGB to luma
float c0_Y = CtL(c[0]);

float luma[25] = float[](c0_Y, CtL(c[1]), CtL(c[2]), CtL(c[3]), CtL(c[4]), CtL(c[5]), CtL(c[6]),
CtL(c[7]), CtL(c[8]), CtL(c[9]), CtL(c[10]), CtL(c[11]), CtL(c[12]),
CtL(c[13]), CtL(c[14]), CtL(c[15]), CtL(c[16]), CtL(c[17]), CtL(c[18]),
CtL(c[19]), CtL(c[20]), CtL(c[21]), CtL(c[22]), CtL(c[23]), CtL(c[24]));

// Precalculated default squared kernel weights
vec3 w1 = vec3(0.5, 1.0, 1.41421356237); // 0.25, 1.0, 2.0
vec3 w2 = vec3(0.86602540378, 1.0, 0.5477225575); // 0.75, 1.0, 0.3

// Transition to a smaller sharpening kernel if the center edge val is above thr
vec3 dW = pow(mix( w1, w2, smoothstep( 0.3, 0.6, edge)), vec3(2.0));

// Partial laplacian outer pixel weighting
float mdiff_c0 = 0.02 + 3*( abs(luma[0]-luma[2]) + abs(luma[0]-luma[4])
+ abs(luma[0]-luma[5]) + abs(luma[0]-luma[7])
+ 0.25*(abs(luma[0]-luma[1]) + abs(luma[0]-luma[3])
+abs(luma[0]-luma[6]) + abs(luma[0]-luma[8])) );

float weights[12] = float[]( ( dW.x ), ( dW.x ), ( dW.x ), ( dW.x ), // c2, // c4, // c5, // c7
( min(mdiff_c0/mdiff(24, 21, 2, 4, 9, 10, 1), dW.y) ), // c1
( min(mdiff_c0/mdiff(23, 18, 5, 2, 9, 11, 3), dW.y) ), // c3
( min(mdiff_c0/mdiff(4, 20, 15, 7, 10, 12, 6), dW.y) ), // c6
( min(mdiff_c0/mdiff(5, 7, 17, 14, 12, 11, 8), dW.y) ), // c8
( min(mdiff_c0/mdiff(2, 24, 23, 22, 1, 3, 9), dW.z) ), // c9
( min(mdiff_c0/mdiff(20, 19, 21, 4, 1, 6, 10), dW.z) ), // c10
( min(mdiff_c0/mdiff(17, 5, 18, 16, 3, 8, 11), dW.z) ), // c11
( min(mdiff_c0/mdiff(13, 15, 7, 14, 6, 8, 12), dW.z) ) ); // c12

weights[4] = (max(max((weights[8] + weights[9])/4, weights[4]), 0.25) + weights[4])/2;
weights[5] = (max(max((weights[8] + weights[10])/4, weights[5]), 0.25) + weights[5])/2;
weights[6] = (max(max((weights[9] + weights[11])/4, weights[6]), 0.25) + weights[6])/2;
weights[7] = (max(max((weights[10] + weights[11])/4, weights[7]), 0.25) + weights[7])/2;

// Calculate the negative part of the laplace kernel
float weightsum = 0;
float neg_laplace = 0;

int order[12] = int[]( 2, 4, 5, 7, 1, 3, 6, 8, 9, 10, 11, 12 );

for (int pix = 0; pix < 12; ++pix)
{
neg_laplace += luma[order[pix]] *(weights[pix]);
weightsum += weights[pix];
}

// Compute sharpening magnitude function, x = edge mag, y = laplace operator mag
float sharpen_val = 0.01 + (curve_height/(curveslope*pow(edge, 3.5) + 0.5))
- (curve_height/(8192*pow((edge*2.4), 4.5) + 0.5));

// Calculate sharpening diff and scale
float sharpdiff = (c0_Y - (neg_laplace/weightsum))*(sharpen_val*0.8);

// Calculate local near min & max, partial cocktail sort (No branching!)
for (int i = 0; i < 3; ++i)
{
for (int i1 = 1+i; i1 < 25-i; ++i1)
{
float temp = luma[i1-1];
luma[i1-1] = min(luma[i1-1], luma[i1]);
luma[i1] = max(temp, luma[i1]);
}

for (int i2 = 23-i; i2 > i; --i2)
{
float temp = luma[i2-1];
luma[i2-1] = min(luma[i2-1], luma[i2]);
luma[i2] = max(temp, luma[i2]);
}
}

float nmax = max(((luma[22] + luma[23]*2 + luma[24])/4), c0_Y);
float nmin = min(((luma[0] + luma[1]*2 + luma[2])/4), c0_Y);

// Calculate tanh scale factor, pos/neg
float nmax_scale = min(((nmax - c0_Y) + L_overshoot), max_scale_lim);
float nmin_scale = min(((c0_Y - nmin) + D_overshoot), max_scale_lim);

// Soft limit sharpening with tanh, lerp to control maximum compression
sharpdiff = mix( (soft_lim(max(sharpdiff, 0), nmax_scale)), max(sharpdiff, 0), L_comp_ratio )
+ mix( -(soft_lim(min(sharpdiff, 0), nmin_scale)), min(sharpdiff, 0), D_comp_ratio );

if (video_level_out == true)
{
if (sharpdiff > 0) { return vec4( orig.rgb + (minim_satloss - c[0].rgb), 1.0 ); }

else { return vec4( (orig.rgb + sharpdiff), 1.0 ); }
}

// Normal path
if (sharpdiff > 0) { return vec4( minim_satloss, 1.0 ); }

else { return vec4( (c[0].rgb + sharpdiff), 1.0 ); }
}


how to convert this to hlsl as 2 pass filter doesn't work for me on mpc.

[BRM]

I've made my port compatible with linear-scaling/sigmoid-upscaling. Also much less distortions now (if any).

Both turned off, mpv spits this out whenever I run it with the 1-pass filter in page 3 (2-pass gist version is unusable now for some reason, slows everything down to a crawl and is overly intense).

[vo/opengl] fragment shader compile log (status=0):
[vo/opengl] 0:96(15): error: type error in array constructor: expected: vec3, found vec4
[vo/opengl]
[vo/opengl] shader link log (status=0): error: linking with uncompiled shader
[vo/opengl] after rendering: OpenGL error INVALID_OPERATION.
[vo/opengl] after rendering: OpenGL error INVALID_OPERATION.
[fpsadjust] Setting speed to 1
AV: 00:00:00 / 00:06:29 (0%) A-V: 0.022 DS: 2.500/0 Cache: 9s+30MB
[vo/opengl] after rendering: OpenGL error INVALID_OPERATION.
[vo/opengl] after rendering: OpenGL error INVALID_OPERATION.
[vo/opengl] after rendering: OpenGL error INVALID_OPERATION.
[vo/opengl] after rendering: OpenGL error INVALID_OPERATION.
AV: 00:00:00 / 00:06:29 (0%) A-V: 0.006 DS: 2.250/0 Cache: 9s+40MB
[vo/opengl] after creating framebuffer texture: OpenGL error INVALID_OPERATION.
AV: 00:00:00 / 00:06:29 (0%) A-V: 0.005 DS: 2.200/0 Cache: 9s+47MB
[vo/opengl] after creating framebuffer texture: OpenGL error INVALID_OPERATION.
AV: 00:00:00 / 00:06:29 (0%) A-V: 0.005 DS: 2.167/0 Cache: 9s+53MB
[vo/opengl] after creating framebuffer texture: OpenGL error INVALID_OPERATION.
AV: 00:00:00 / 00:06:29 (0%) A-V: 0.003 DS: 2.286/0 Cache: 9s+58MB
[vo/opengl] after creating framebuffer texture: OpenGL error INVALID_OPERATION.
AV: 00:00:00 / 00:06:29 (0%) A-V: 0.001 DS: 2.250/0 Cache: 9s+67MB
[vo/opengl] after creating framebuffer texture: OpenGL error INVALID_OPERATION.
AV: 00:00:02 / 00:06:29 (0%) A-V: -0.005 DS: 2.389/0 Cache: 9s+302MB

[ts1]

Post whole log.

Try updated.

[BRM]

Post whole log.

Try updated.

Too strong and slows down video playback a LOT (literally cannot run at more than 3fps, using a gigabyte 280x with the radeon git driver).

The rest of the log contained the shader and that's it.

[ts1]

It's a pretty lightweight shader. Something wrong with your GPU drivers/setup. I tried it on Linux and it works fine with proprietary nVidia drivers, but with nouveau it's also not usable, very slow.
There is parameter to control sharpening strength.

[BRM]

It's a pretty lightweight shader. Something wrong with your GPU drivers/setup. I tried it on Linux and it works fine with proprietary nVidia drivers, but with nouveau it's also not usable, very slow.
There is parameter to control sharpening strength.

I know, I get similar results with the 2-pass shader.
As for my setup, it's radeon-git, xorg.conf and xorg.conf.d/20-radeon.conf are as follows: http://pastebin.com/raw/aE3e21Bn

Section "Device"
Identifier "radeonVGA"
Driver "radeon"
Option "AccelMethod" "Glamor" # See compatibility on features page
Option "DRI3" "on" # Compatibility not very well documented, remove it if it fails
Option "EnablePageFlip" "On"
EndSection

[ts1]

Can't help you with this, I never used Radeon and I use Linux (Sabayon, Gentoo based) from flash drive, it already includes proprietary drivers and mpv.

[XMonarchY]

How would I go about integrating your latest version downloads into ReShade? ReShade uses either .fx files or .h files. I assume I can just rename TWO-PASS .hlsl file to .h or .fx file because the content is identical, isn't it? When I open the file with WordPad, it shows identical info. I assume ReShade uses the TWO-PASS file because ATM it uses a 12KB file, which is closer to TWO-PASS 11KB then to ONE-PASS 4KB file.

[aufkrawall]

You have to port the HLSL shader to ReShade shading language.
bacondither or JPulowski will do it soon:
https://github.com/crosire/reshade-shaders/issues/8

[BRM]

Can't help you with this, I never used Radeon and I use Linux (Sabayon, Gentoo based) from flash drive, it already includes proprietary drivers and mpv.

New 1 pass shader causes a segfault.

LLVM triggered Diagnostic Handler: SIInstrInfo::storeRegToStackSlot - Can't spill VGPR!
[1] 9181 segmentation fault mpv https://www.youtube.com/watch\?v\=yqynPcFEL3U

[ts1]

Are you sure this is shader's fault? Is there any errors from a shader compiler?
I will test it with mesa later.

Edit: works with mesa, slow but without errors.
LLVM triggered Diagnostic Handler: SIInstrInfo::storeRegToStackSlot - Can't spill VGPR!
Seems like it's a LLVM's bug https://bugs.freedesktop.org/show_bug.cgi?id=75276#c50. Do you use latest LLVM 3.7?

[BRM]

Seems like it's a LLVM's bug https://bugs.freedesktop.org/show_bug.cgi?id=75276#c50. Do you use latest LLVM 3.7?

Yes, that one is super slow but works, and you were right, it was caused by using LLVM 3.5.

[ts1]

nevermind

[ts1]

bacondither
I removed my previous comment, but there is still a problem in minim_satloss, there are cases when c0_Y==0.

Ok, this is only in glsl so.

[foozoor]

Thanks for the update ts1. ;)

[XMonarchY]

Could someone provide some images that compare LumaSharpen, AdaptiveSharpen, FineSharp, and or Crispen Edges using similar settings?

I usually apply a little bit of AdaptiveSharpen (0.1-0.15) and a little bit of LumaSharpen (0.50 - strength, 0.030 - clamp) in films and games, but I'd like to reduce ringing, noise, artifacts, overshoot, and other side-effects. What should I be using them - both AdaptiveSharpen and LumaSharpen OR just one of them? Which one?

[bacondither]

New version released! :cool:

bacondither
I removed my previous comment, but there is still a problem in minim_satloss, there are cases when c0_Y==0.

Ok, this is only in glsl so.

That's strange, i have done a glsl port of my own for testing so i could poke around with it in mpv a bit.

[toniash]

A guy (igv) ported and improved this shader for the mpv hook system:
https://gist.github.com/igv/8a77e4eb8276753b54bb94c1c50c317e

I don't really know what he improved but it looks amazingly on my sister's laptop with a weak intel hd3000 iGPU. The two-passes version is too intensive!

(From Madvr thread)

[bob88]

Hello,

I'm using latest version MPC-HC x64 1.7.10
I added the shaders in post-resize in the described order but I still get a green screen.
Could somebody please tell me how to fix it?

Thank you!

[bacondither]

Could somebody please tell me how to fix it?

Thank you!

If you are using madVR as video renderer untick "lose BTB and WTW if it improves performance" and "run custom pixel shaders in video levels instead of PC levels".
The settings can be found in madVR settings, rendering->trade quality for performance.

[bob88]

I am using the default 'enhanced video renderer (custom presenter)'
madVR is listed as unavailable, I guess it has to be installed additionally. Should I be using that? Is it better?

[bacondither]

Ok you are using EVR.
Go to View-> Renderer Settings -> Presentation and enable "Full floating point point processing" or "Half floating point point processing", then restart MPC-HC.

Hmm i should probably add this to the installation instructions.

[BetA13]

Ok you are using EVR.
Go to View-> Renderer Settings -> Presentation and enable "Full floating point point processing" or "Half floating point point processing", then restart MPC-HC.

Hmm i should probably add this to the installation instructions.

ha, and i was wondering why it didnt work in my normal MPCHC test player but did work with madvr etc..

Thanks for the tip with the "Full floating point point processing"


Just wondering, what does this function do anyway?

"Full floating point point processing", "Half floating point point processing"


Have a nice evening bacondither and thanks for teh lesson haha..

[bob88]

Ok you are using EVR.
Go to View-> Renderer Settings -> Presentation and enable "Full floating point point processing" or "Half floating point point processing", then restart MPC-HC.

Hmm i should probably add this to the installation instructions.

Enabling full or half floating point processing just results in completely black screen regardless of shaders.

[shaolin95]

A guy (igv) ported and improved this shader for the mpv hook system:
https://gist.github.com/igv/8a77e4eb8276753b54bb94c1c50c317e

I don't really know what he improved but it looks amazingly on my sister's laptop with a weak intel hd3000 iGPU. The two-passes version is too intensive!

(From Madvr thread)
Where do you place this file to work with mpv (SMPlayer)? I have been using it with MPC-HC but wanted to try mpv as it seems to be better for SVP.
Thanks

[toniash]

Where do you place this file to work with mpv (SMPlayer)? I have been using it with MPC-HC but wanted to try mpv as it seems to be better for SVP.
Thanks


vo=opengl-hq:backend=angle:scale=ewa_lanczossharp:cscale=ewa_lanczossoft:interpolation:tscale=mitchell:tscale-clamp:temporal-dither:user-shaders="~~/shaders/adaptive-sharpen.glsl"

[shaolin95]

vo=opengl-hq:backend=angle:scale=ewa_lanczossharp:cscale=ewa_lanczossoft:interpolation:tscale=mitchell:tscale-clamp:temporal-dither:user-shaders="~~/shaders/adaptive-sharpen.glsl"

Sorry for the dumb question but is my first time using mpv.
Will I add in the same section I add this line for SVP "--input-ipc-server=mpvpipe" as shown on this guide?

https://www.svp-team.com/wiki/SVP:SMPlayer

[toniash]

Sorry for the dumb question but is my first time using mpv.
Will I add in the same section I add this line for SVP "--input-ipc-server=mpvpipe" as shown on this guide?

https://www.svp-team.com/wiki/SVP:SMPlayer

yes:):)

[shaolin95]

yes:):)
I must be doing something wrong cause I just get an error. I was dropping the script file in the root directory of the mpv64 folder (inside of the SVP folder).
Is that what I am doing wrong? :)

[toniash]

I must be doing something wrong cause I just get an error. I was dropping the script file in the root directory of the mpv64 folder (inside of the SVP folder).
Is that what I am doing wrong? :)

Then I think you need "user-shaders="~~/adaptive-sharpen.glsl"

Good luck!

[toniash]

Then I think you need "user-shaders="~~/adaptive-sharpen.glsl"

Good luck!

Did it work?:confused:

[shaolin95]

Did it work?:confused:

I didn't get an error but I do not see any effect on the screen so is not working :(
Thanks though, I just didn't want to keep bugging you. :)

[toniash]

I didn't get an error but I do not see any effect on the screen so is not working :(
Thanks though, I just didn't want to keep bugging you. :)

#define curve_height .96
// Main sharpening strength, POSITIVE VALUE ONLY! // 0.3 <-> 1.5 is a reasonable range of values

what value do you use?

[shaolin95]

#define curve_height .96
// Main sharpening strength, POSITIVE VALUE ONLY! // 0.3 <-> 1.5 is a reasonable range of values

what value do you use?
I didn't change it at all just let the script untouched. I'm just not sure I'm putting the file in the right folder maybe.
I will take some screenshots of my config and post.
Thanks!