need help witch the plugin RemapFrames

[gil900]

ReplaceFramesMC(BaseClip,142,7)
# reinterlace the video
AssumeTFF()
SeparateFields()
Weave()

I don't see a de-interlace routine in your script. The last two lines above do unnecessary work. They don't accomplish anything.

Okay.

It works but does not work if there is more than one ReplaceFramesMC() line/

I mean -
this works:

...}

ReplaceFramesMC(BaseClip,142,7)

# reinterlace the video
AssumeTFF()
SeparateFields()
Weave()

But if I will add another row of ReplaceFramesMC() then nothing of ReplaceFramesMC will works including the - ReplaceFramesMC(BaseClip,142,7).

I do not know why ..

[manono]

RX(142,7)
RX(FirstFrameNumber,NumberOfFramesToBeReplaced)
.
.
.

[sanlyn]

But if I will add another row of ReplaceFramesMC() then nothing of ReplaceFramesMC will works including the - ReplaceFramesMC(BaseClip,142,7).

Try This (I added fictitious numbers in ReplaceFramesMC):

BaseClip=Mpeg2Source("part 1.d2v")
BaseClip=BaseClip.RemapFrames(mappings="
[341 346] [154 159]
[571 575] [384 388]
", SourceClip=Mpeg2Source("part 1 v2(r1).d2v"))

# -- make last version of BaseClip the current focus ---
BaseClip
ReplaceFramesMC(142,7)
ReplaceFramesMC(77,2)
ReplaceFramesMC(88,1)
ReplaceFramesMC(99,1)

return last

function ReplaceFramesMC(clip Source, int N, int X)
{
    # N is number of the 1st frame in Source that needs replacing. 
    # X is total number of frames to replace
    #e.g. RX(101, 5) would replace 101,102,103,104,105 , by using 100 and 106 as reference points for mflowfps interpolation
 
    start=Source.trim(N-1,-1) #one good frame before, used for interpolation reference point
    end=Source.trim(N+X,-1) #one good frame after, used for interpolation reference point
 
    start+end
    AssumeFPS(1) #temporarily FPS=1 to use mflowfps
  
    super = MSuper()
    backward_vec = MAnalyse(super, isb = true)
    forward_vec = MAnalyse(super, isb = false)
    MFlowFps(super, backward_vec, forward_vec, blend=false, num=X+1, den=1) #num=X+1
    AssumeFPS(FrameRate(Source)) #return back to normal source framerate for joining
    Trim(1, framecount-1) #trim ends, leaving replacement frames
  
    Source.trim(0,-N) ++ last ++ Source.trim(N+X+1,0)
}

When you don't include the specific clip name in RemoveFramesMC() or in most other functions, the "clip" is usually assumed to be the "last" entity that was referenced.

[gil900]

But if I will add another row of ReplaceFramesMC() then nothing of ReplaceFramesMC will works including the - ReplaceFramesMC(BaseClip,142,7).

Try This (I added fictitious numbers in ReplaceFramesMC):

BaseClip=Mpeg2Source("part 1.d2v")
BaseClip=BaseClip.RemapFrames(mappings="
[341 346] [154 159]
[571 575] [384 388]
", SourceClip=Mpeg2Source("part 1 v2(r1).d2v"))

# -- make last version of BaseClip the current focus ---
BaseClip
ReplaceFramesMC(142,7)
ReplaceFramesMC(77,2)
ReplaceFramesMC(88,1)
ReplaceFramesMC(99,1)

return last

function ReplaceFramesMC(clip Source, int N, int X)
{
    # N is number of the 1st frame in Source that needs replacing. 
    # X is total number of frames to replace
    #e.g. RX(101, 5) would replace 101,102,103,104,105 , by using 100 and 106 as reference points for mflowfps interpolation
 
    start=Source.trim(N-1,-1) #one good frame before, used for interpolation reference point
    end=Source.trim(N+X,-1) #one good frame after, used for interpolation reference point
 
    start+end
    AssumeFPS(1) #temporarily FPS=1 to use mflowfps
  
    super = MSuper()
    backward_vec = MAnalyse(super, isb = true)
    forward_vec = MAnalyse(super, isb = false)
    MFlowFps(super, backward_vec, forward_vec, blend=false, num=X+1, den=1) #num=X+1
    AssumeFPS(FrameRate(Source)) #return back to normal source framerate for joining
    Trim(1, framecount-1) #trim ends, leaving replacement frames
  
    Source.trim(0,-N) ++ last ++ Source.trim(N+X+1,0)
}

When you don't include the specific clip name in RemoveFramesMC() or in most other functions, the "clip" is usually assumed to be the "last" entity that was referenced.

Still the same problem.

No matter .. I just created a new avi file (with patches of RemapFrames) called setep 2.avi then I worked on it with ReplaceFramesMC.

I converted with "lagarith lossless codec" to YUY2 colorspace because I had to .. (ReplaceFramesMC not work with RGB)

[sanlyn]

Odd that it didn't work. I ran it on 3 different files -- it's an old script like the one that I used in the previous thread from a few weeks ago, but tested again just in case. Anyway, I'm glad you made it work

[gil900]

Can you give me a simple guide to color repair?

I change the colors by what I see. But I noticed that it is impossible to find color settings which fit to all the tape.

The question is if this happens because I'm not professional enough to get perfect settings (suitable to the tape) or is it a problem that even professionals have trouble to solve.

Is there a way to set color settings not only by what i see these by some technical information?

Look at this one for example:


I know the color of the cabinets is yellow, not green.
So I played with the color settings to get yellow color.

But, I'm not sure I totally correct that the right color is yellow.

Is there a way to check the point? Way to know for sure what the correct color of the cabinets?
Maybe 20 years ago was really the color was green and I do not remember ..

This is a problem

[sanlyn]

Even pros have problems, especially with old tape. Especially with damaged tape. It's not simple -- but it's not impossible. I often wonder why a beginning guide to color isn't posted here. I leaned some of it early in college film class, but most of it from internet tutorials about color correction with Photoshop Pro.

Color correction by eyeball alone makes it more difficult -- yet, ultimately, it's what you see that tells you whether it's "right' or "wrong". The problem is knowing why things don't look correct. With some experience you can make educated guesses. But when you're starting to put basic principles together you are lost unless you have a way to read the color values of pixels in an image. Pixel readers (sometimes called "color pickers" or "color samplers") are part of the software controls in high-end graphics applications. But a free one that sits on your desktop and can be used to read pixel colors in VirtualDub with the mouse pointer is here: http://www.netreach.net/~gavin/gavsfreeware/csamp.htm .

A pixel reader can tell you why the whites aren't white, the grays aren't gray, and/or the blacks aren't really black, and why. A very dark gray that is almost black would have RGB pixel values of R32 G32 B32. A middle gray (such as the gray shadow and folds in a white shirt) would be about R128 G128 B128. The brightest part of a white shirt or piece of paper would be about R200 G200 B200, perhaps a little brighter or dimmer depending on the lighting. You'll notice that the RGB values in the dark gray, mid-gray, and white are made of equal values for R, G, and B. If you can get gray and white objects in an image close to those hues, then R, G, and B will more in balance and other colors fall into place. Skin color without shadow or bright highlights would be R120 G90 B65, having various lighter or darker values but in approximately those proportions of R, G, and B.

So, even if you can read pixel values you have to understand what you're reading. Many people learn, but it's not an overnight project. I don't know what you are using for color controls, but working in RGB seems to be easier and more intuitive for people to understand. An easy color filter to use in VirtualDub is ColorMill ( http://fdump.narod.ru/rgb.htm ). If you are working in YUV, it gets more complicated because changing one color changes the other two. In RGB, you can adjust color channels individually.

Keep in mind that VHS color is corrupt and uneven anyway. On damaged media, it's worse. Often there is no way to correct it completely. It's best to start by adjusting darks, mids and brights so that they look natural (with ColorMill you use its Dark, Mid, and Bright Level controls for that). Then you adjust color in those same ranges. Color adjustments can affect contrast, so you do have to fiddle back and forth between Levels and Colors for a while.

No, on most tapes there is no single adjustment for the entire video. That's just the perverse nature of analog, especially tape. For the time being, I'd keep it simple and do the best you can. Be patient, and put a few hours into trying something like ColroMill to get a feel for color and level effects. But I think you'll find no quick fix for this area. Pros and serious hobbyists break videos into many scene-by-scene correction projects. I think that would be demanding for you at this point, and most people wouldn't go that far.

Now you know why it takes lots of money and months or years of effort to correct old Hollywood movies for digital video.

[poisondeathray]

I know the color of the cabinets is yellow, not green.
So I played with the color settings to get yellow color.

But, I'm not sure I totally correct that the right color is yellow.

Is there a way to check the point? Way to know for sure what the correct color of the cabinets?
Maybe 20 years ago was really the color was green and I do not remember ..

This is a big problem

There is no way to know for sure unless you have known objects shot at the same time (and under the same conditions, like lighting) e.g. a color chart is often used at a beginning of a shoot, but it's unlikely that this would happen in a consumer home video . For something like this you often have to "eyeball" it or guess what the approximate color is in relation to everything else

[sanlyn]

It isn't always possible to match the exact "real" color of very object in a home video, especially with home lighting and mixed light sources (for example, a small lamp on one side of the room but a bright neon light on the other side). The more workable approach (and the one that pros often take) is to make skin tones look natural, or close to it. Or, make sure the family pet is at least recognizable. If you make the cabinets the exact yellow, but your children look purple, it would be better to make people look convincing.

Hollywood films and professional photos can be photographed in a few minutes or seconds; but the pros spend all day on the lighting, reflector arrays, test exposures, re-shootings, and all that. Home video doesn't have those luxuries. It is possible to make almost everything look "proper" -- but it takes time, and you'd need very sophisticated and difficult-to-learn color controls, and you'd try techniques such as contrast and layer masking. It would take forever. I get weary just thinking about it.

[gil900]

I want people think I filmed all this yesterday

Pros and serious hobbyists split videos into many scene-by-scene correction projects. I think that would be demanding for you at this point, and most people wouldn't go that far.

I will certainly do so. I've been doing it originally. But for the colors I'll have to cut into smaller pieces of video. It was not the original plan ..

[sanlyn]

Old tape has many problems. One of them is uneven fading of color information (reds might fade faster than blue or green), or uneven color stains and blotches along edges or borders. Sometimes you might notice that most viewers focus on the people in family videos; they don't even notice the furniture, and they seldom recall or even notice colors. In the image of the kitchen you posted earlier, I noticed a portion of a wall below the cabinets and an area of white above them. If that wall and upper area are shades of whites and grays, they are guides to the overall color balance. Grays and whites should have equal portions (or approximately equal) of Red, Green, and Blue., whether they are light or dark in intyenbsity.

In the images you posted, the "before" picture looks more accurate to me than the "after". The left-hand cabinets are closer to yellow than are the right-hand (after) cabinets; the left cabinets look closer to orange than to yellow. I notice that the cabinet on the right-hand side of both images has a dark fixture along its bottom border. In the "before" image that object is nearly a black color, and below it and in the right corner there is something green (a plant?). In the "after" image, the black object and cabinet knobs are closer to dark brown; the plant is losing its green color (well, I think that's a plant ? ?). Also, look at the bright, upper portion in both images. They look stained (magenta or pink areas, sometimes uneven blue).

If you use a pixel reader, you can see these color values. The cabinets in the left image have average pixel values of R175 G165 B95. The cabinets in the right ("after") image in the brighter areas have average pixel values of R235 G190 B120. Those numbers tell me that Red has been added to the original, and the later images are brighter overall. If you know basic color theory, you would know that Yellow is a "secondary" color, so called because it is made by mixing two primary colors (Red + Green). A perfect yellow would have RGB values such as R200 G200 B0. But there are few "pure" colors in nature, not even in a blue sky. So when you say the cabinets in real life are "yellow", a colorist would ask "what shade of yellow?" and "what kind of lighting is in the room?" To me, the cabinets seem to have some wood grain rather than paint, so I would say the color you're looking for is somewhere between the greener left-hand image and the more red right-hand image.

Brightening midtones in the left-hand image makes a difference in how you see some of the colors. Also, in both images there is too much bright blue, but there is less blue in the midtones and darks. That isn't unusual, especially for old tape with uneven color. A filter like ColorMill lets you correct colors and levels in 3 ranges, so you cold lower bright blue with very little effect on the lower blue. If the walls in the lower part of the image are supposed to be shades of white or light gray (the shadowed areas will be shades of gray, not bright white), then I would read some pixels to see if there is a major RGB imbalance in those areas. (Pixel readers read pixels in groups of 3, so you see the average of a group of pixels, not just a single pixel. It is impossible for every pixel in an area to have exactly the same RGB value). Keep your eye on other objects as well; I see objects of various colors inside the cabinets, and those can act as guides as well.

In any case, the colors in the right-hand "after" image look closer to what I think you want. The images do look "stained" in certain areas, so don't expect perfection.

[gil900]

From what you say I have to go by the rule:
"Do not try to invent new colors"

Is that what you recommend?


EDIT

I found a proof and a sign that justifies the action did - adding red.

Take a look:



Where did circle, you can clearly see that there is a surplus of blue.
This blue color is not natural of course. And the opposite color to blue is red.

So I guess I made a logical action (add red) ..
I never thought of this mark before. I relied on memory ...

What do you think?

Maybe I was wrong that I added too much red.
But in that case is there a way to know how many red added by this sign?

[gil900]

Okay.
An update ..

Look what I did.

The basic assumption is that the color blue over the coffin is not natural.
So I thought the blue color disrupts the color of the cabinets.

So what I did was I calculated the color is supposed to be ..

And I did it this way -

I took the unnatural blue color:


and i got:
R=215
G=220
B=230

Then I took the color should be (not blue) from this point:


I got:
R = 255
G = 255
B = 255

Based on this, I knew that in order to fix the blue part I should add for R - 40 (255-215), for G - 35 (255-220) and for B - 25 (255-230)

so the color from the Cabinets is:



R = 155
G = 135
B = 75

So to correct the color I need to add for R - 40 , for G - 35 and for B - 25

The result:
R = 195
G = 175
B = 100


The final result:



Is that correct?

[jagabo]

I suspect that blue tinge is only at the top of the frame.

[sanlyn]

I think jagabo is correct. Judging from your earlier videos, the tape has visible rainbows and stain across the top of the images. It's easier to see against bright colors such as white. It's not possible to fully correct uneven discolorations such as those, although correcting for overall color balance and smoothing-out "peaks" in certain color ranges can sometimes make rainbows and stains less visible -- as you can see in the above images. There are also some Avisynth plugins that can mitigate the problem.

The video has other color problems that Avisynth techniques might help. For example, look at the cabinet doors in post #43, bottom photo. In one of the tall doors near the image center, in that door's lower right corner, there is a blotch of excess blue. I saw these discolorations appear and disappear at random times throughout the video. This is usually caused by storing tape is overly warm or humid conditions, but many other causes have been cited. The blue patches are tape damage. Much of the time, these discolorations pass so quickly at the playing speed that often they are not noticed.

[jagabo]

The discolored top of the frame could be fixed differently than the rest of the frame by creating two filtered streams and blending them with an alpha mask.

[gil900]

The discolored top of the frame could be fixed differently than the rest of the frame by creating two filtered streams and blending them with an alpha mask.

Looks like the next thing I had to learn as RemapFrames.

Thank you.
I'll look about it on Google and the forum And if I will not understand so I ask ..

[jagabo]

An example of using a mask to filter only a section of the frame:
https://forum.videohelp.com/threads/349806-Help-removing-a-fixed-white-spot-on-all-my-v...=1#post2191263

General explanation of alpha masking:
https://forum.videohelp.com/threads/347583-Color-correction-%28or-grading-%29?p=2174838...=1#post2174838

[sanlyn]

And the opposite color to blue is red.

The opposite to blue isn't red. Its opposite is Yellow (red+green).

If object is too blue, then subtract blue or add yellow (red+green)
If object is too red, then subtract red or add cyan (blue + green)
If object is too green, then subtract green or add magenta (blue+red)

[gil900]

And the opposite color to blue is red.

The opposite to blue isn't red. Its opposite is Yellow (red+green).

If object is too blue, then subtract blue or add yellow (red+green)
If object is too red, then subtract red or add cyan (blue + green)
If object is too green, then subtract green or add magenta (blue+red)

Okay.
It's very new for me ..

I guess I have to do a lot of homework ...

Anyway, I think I do not have to do all these things ..
In this post:
https://forum.videohelp.com/threads/348054-Help-with-filtering-pictures-problems-from-V...=1#post2178028

I reread again your post 69.
In the past, it seems I did not do your instructions (step 1.avs, step 2.avs and step 3.avs). why I do not remember.
Seems I had some mistakes and I was not experienced enough to solve them ..
And I did not understand your instructions within the scripts.


Just yesterday I tried your instructions seriously for the first time and I succeeded create the example1.avi just like you did.

And I was very surprised at the difference between the result without your orders, and the result with your instructions.
I'm not even dreamed of such a result ..

So thank you very much!

I had to add another step just to RemapFrames (step 0.avs) but everything else go according to your instructions.

[Isac.martins]

I don't use it to fix broken frames, but use it to filter different sets of frames in different ways. For this I use the ReplaceFramesSimple part of it. And use it something like this (from a script I happen to have lying around):

BaseClip=Last
SourceClip=BaseClip
SourceClip=SourceClip.Tweak(Bright=0,Cont=0.93,Cor ing=False)
ReplaceFramesSimple(BaseClip,SourceClip,Mappings="[6285 7433] [10234,12444] [15023 18400] ")

BaseClip=Last
SourceClip=BaseClip
SourceClip=SourceClip.InpaintFunc(mask="logo1.bmp" ,loc="22,18,-236,-244",AR=4.0/3.0,mode="Inpaint",speed=10, ppmode=2,pp=85,radius=6.0,preblur=10.0)
ReplaceFramesSimple(BaseClip,SourceClip,Mappings="[3695 208691]")

The other parts of RemapFrames are used similarly (I think). But if you're really talking about fixing frames, as in generating new frames to replace corrupted or ugly ones, then RemapFrames isn't what you want.

''But if you're really talking about fixing frames, as in generating new frames to replace corrupted or ugly ones, then RemapFrames isn't what you want''
I want this goal, can you help me with this role?

[johnmeyer]

You would want a script like this one I developed several years ago:

Finding individual "bad" frames in video; save frame number; or repair

Note that in that first post I provided a link to a better version of the script which I posted later in the thread (post #6).