I need help in filters to improve quality

[JoJoJoJo]

Hello

I need the filters settings that make photo number 1 same photo number 2
I use MeGUI and StaxRip to encode the video

Photo number 1 :

https://forum.videohelp.com/images/imgfiles/PJsJ6mW.jpg


Photo number 2 :

https://forum.videohelp.com/images/imgfiles/3Ym2oMO.png

[jagabo]

Ouch. That's insane. Did he use too much tanning cream? But if you must, somewhere around Tweak(sat=2.3, hue=-8, cont=1.33, bright=-10).

[budwzr]

This is what Vegas comes up with as a "color match"

[sanlyn]

I'm afraid we need a little information here. Where did the top "good" image come from? Where did the ugly bottom image
come from? How are you processing these videos? How did the "bad" image get so screwed up? Why do you use huge
bold multi-colored screaming fonts?

I don't know that anyone can make both images "look alike" exactly. It appears that Vegas tries to create an "averaging"
histogram, which is not the way it's done -- especially when the two images are so different. In the "corrected" version
above, saturation far exceeds that in the original and both people appear to have terminal liver disorders.

I'm not sure what you mean by "make this one look like that one", especially when the "good" copy looks a lot better but
could still use some work. Maybe it depends on the scene''s mood; the woman looks great, but the guy looks as if he
needs a good meal. I don't know, maybe the scene is supposed to look that way. In any case I don't think you'll
get an exact match.

I'll explain what I did below, even if you won't understand a word of it and would never attempt such a thing yourself.
You probably won't like it, because I don't think you really want the bad one to look like the good one. I think you really
want something else, because most "average" viewers would prefer something closer to the badly distorted
bottom image. But we'll see...

First, begin by assuming that magic buttons won't work. Next, go on at a later time to get familiar with basic color
theory, valid video luma and chroma levels, learn what gross over saturation looks like, learn what over juiced ugly
contrast looks like, then learn to use RGB color controls, pixel value readers. and histograms. If you don't learn to use
any of this stuff, you'll never find an "auto" filter that will work on ugly distorted screwed-up video for you.

Below, the original image (you couldn't get a small video sample for us?):

[Attachment 23407 - Click to enlarge]

Above, the original is badly oversaturated, darks are dense and very nearly crushed out of existence by pumping
contrast like a bandit.

Below, Step 1 was to reduce saturation by 33% with the ColorMill filter. This allows one to see
immediately that the original is too green, even if the woman's blue scarf glows like neon and bleeds off the screen:

[Attachment 23408 - Click to enlarge]

Below: Steps 2, 3 and 4 used three different layers of VirtualDub's gradation curves. Each layer is a tweak of the one
above it. Not an exact match: the scarf is a little too blue, my version is a bit brighter than the original "good" image,
and there's more blue in the walls. At least the people don't look ill or freaky. They seem to look human. None of
this work does anything about the oversharpening or the banding and clumpy-grain compression artifacts in the
"bad" image. Those will look far worse when the final video is encoded.

[Attachment 23409 - Click to enlarge]

The filters used were ColorMill and three chained copies of the gradation curve plugin. To determine their filter
values I first took pixel readings from vital areas in the "good" version of the posted images. The "bad" image was
mounting in AfterEffects, where I used a combination of ColorFinesse color wheels and curve filters until I could
make most areas have the same or near pixel values that I had taken from the "good" image. I then used
the ColorFinesse white level and black level controls to tweak those areas so that blacks looked black and
skin tones looked a little more healthy. I then translated those filter values into the VirtualDub versions of
those same filters, which are ColorMill.vdf and Gradation.vdf. Those two plugins and a free pixel sampler
desktop app called Csamp are included in the attached .zip file. I also included a .vcf file from VirtualDub that
contains the filter configuration and settings that I used for the new image. If you have the .vdf files loaded
in your VirtualDub plugins folder, you can use that .vcf file to load those filters and the "bad" image directly
into VirtualDub. This will allow you to see how the filters were set up in detail. With the image showing in
VirtualDub, you can turn the filters off and on, one by one, to see their effect.

One doesn't really need ColorFinesse for this work, it just helps to speed up the process. The pixel reader
tells you the values that you basically want to correct for: you just slide the filter controls to get those effects.
Work on the brights, then the darks, then the middle values. If you don't know what any of those terms
mean, then....well, you really should know this stuff if you want to improve your videos. Otherwise, live with
whatever comes out of the box. I takes time, I know. Watching it, you can learn a lot.

[sanlyn]

I'm afraid we need a little information here. Where did the top "good" image come from? Where did the ugly bottom image
come from? How are you processing these videos? How did the "bad" image get so screwed up? Why do you use huge
bold multi-colored screaming fonts?

I don't know that anyone can make both images "look alike" exactly. It appears that Vegas tries to create an "averaging"
histogram, which is not the way it's done -- especially when the two images are so different. In the "corrected" version
above, saturation far exceeds that in the original and both people appear to have terminal liver disorders.

I'm not sure what you mean by "make this one look like that one", especially when the "good" copy looks a lot better but
could still use some work. Maybe it depends on the scene''s mood; the woman looks great, but the guy looks as if he
needs a good meal. I don't know, maybe the scene is supposed to look that way. In any case I don't think you'll
get an exact match.

I'll explain what I did below, even if you won't understand a word of it and would never attempt such a thing yourself.
You probably won't like it, because I don't think you really want the bad one to look like the good one. I think you really
want something else, because most "average" viewers would prefer something closer to the badly distorted
bottom image. But we'll see...

First, begin by assuming that magic buttons won't work. Next, go on at a later time to get familiar with basic color
theory, valid video luma and chroma levels, learn what gross over saturation looks like, learn what over juiced ugly
contrast looks like, then learn to use RGB color controls, pixel value readers. and histograms. If you don't learn to use
any of this stuff, you'll never find an "auto" filter that will work on ugly distorted screwed-up video for you.

Below, the original image. It's over saturated, with dense dark nearly crushed nelow zero-black and contrast
pumped-up like a bandit.


[Attachment 23416 - Click to enlarge]

Below, Step 1 was to reduce saturation by 33% with the ColorMill filter. This allows one to see
immediately that the original is too green, even if the woman's blue scarf glows like neon and bleeds off the screen:

[Attachment 23417 - Click to enlarge]

Below: Steps 2, 3 and 4 used three layers of VirtualDub's gradation curves. Each layer is a tweak of the one
above it. Not an exact match: the scarf is a little too blue, my version is a bit brighter than the original "good" image,
and there's more blue in the walls. At least the people don't look ill or freaky. They seem to look human. None of
this work does anything about the oversharpening or the banding and clumpy-grain compression artifacts in the
"bad" image. Those will look far worse when the final video is encoded.(Left-click to enlarge)

[Attachment 23418 - Click to enlarge]

The filters used were ColorMill and three chained copies of the gradation curve plugin. To determine their filter
values I first took pixel readings from vital areas in the "good" version of the posted images. The "bad" image was
mounting in AfterEffects, where I used a combination of ColorFinesse color wheels and curve filters until I could
make most areas have the same or near pixel values that I had taken from the "good" image. I then used
the ColorFinesse white level and black level controls to tweak those areas so that blacks looked black and
skin tones looked a little more healthy. I then translated those filter values into the VirtualDub versions of
those same filters, which are ColorMill.vdf and Gradation.vdf. Those two plugins and a free pixel sampler
desktop app called Csamp are included in the attached .zip file. I also included a .vcf file from VirtualDub that
contains the filter configuration and settings that I used for the new image. If you have the .vdf files loaded
in your VirtualDub plugins folder, you can use that .vcf file to load those filters and the "bad" image directly
into VirtualDub. This will allow you to see how the filters were set up in detail. With the image showing in
VirtualDub, you can turn the filters off and on, one by one, to see their effect. Note that most images in the
forum viewer appear darker than they do in any other viewer I'm familiar with.

One doesn't really need ColorFinesse for this work, it just helps to speed up the process. The pixel reader
tells you the values that you basically want to correct for: you just slide the filter controls to get those effects.
Work on the brights, then the darks, then the middle values. If you don't know what any of those terms
mean, then....well, you really should know this stuff if you want to improve your videos. Otherwise, live with
whatever comes out of the box. I takes time, I know.

[Selur]

@sanyln: May be I'm missing something, but aren't you adjusting the wrong image?
From the looks of it he wanted help to adjust the normal looking image to an over-saturated one.
From a quick look it looks like you are adjusting the over-saturated one to look more normal.

[sanlyn]

That's what I think I said JoJoJoJo wanted. Anyway I'm too tired to fix the typing because of the O.P.'s huge
JPG that overfills the page on my laptop screen. The O.P. can use any contrast and saturation filter to make all of
his videos as ugly as he wants. What's so complicated? That's the wonder of digital processing; it's so easy to
make your movies look uglier than VHS could ever hope to be. The trick for those of us who aren't blind is fixing
what the half-blind are doing. I understand that ugly is what many people want. I just don't understand why. Perhaps
they are afflicted with visual disorders ? ? ? JoJoJoJo never answered my question of how the ugly one was obtained.
If he already has the ugly one, why deface #1 to look like it? Perhaps he has entered some sort of contest.

I'd put a smiley here but I have to scroll all the way to the other side of this stretched window to get there. I'll
just hit Alt+S. I don't know the keyboard shortcut for a smiley.

[lordsmurf]

#2 is also missing details. It's softer.
Lots of posterizing too.

FYI: Big fonts and big images is a turn-off to most forum users. (Me included.)

[JoJoJoJo]

Sorry for delay in reply because I have some problem prevented me to reply

Thank you all for your help