Color Correcting the Dragon Box - 3 Part Spectacular

[HakkaiBills93]

No it's the main reason why i gave up doing cc i won't be satisfied with what i do if it isn't perfect and it's impossible to do that alone

[tellyzbad1]

[tellyzbad1]

i didn't wanted to color match each frame but correct them manually as it's not so hard to color correct , the color cast is the main things that can change in some part so if you know how to remove color cast from one frame, just did it frame by frame is easy.
The main problem when you color correct a video is that you have sometimes differrent color cast on some parts so settings can't work on the full episode

There are on average 35,000 frames per episode and you want to manually correct them frame by frame? Are you serious?

He probably thinks film flickering requires manual balancing for it to be a good color correction :/

[HakkaiBills93]

i didn't wanted to color match each frame but correct them manually as it's not so hard to color correct , the color cast is the main things that can change in some part so if you know how to remove color cast from one frame, just did it frame by frame is easy.
The main problem when you color correct a video is that you have sometimes differrent color cast on some parts so settings can't work on the full episode

There are on average 35,000 frames per episode and you want to manually correct them frame by frame? Are you serious?

He probably thinks film flickering requires manual balancing for it to be a good color correction :/

no some part in the same reels have really differrent colors (even color cast is differrent) you can do a full part perfectly for those episodes with the same settings?

[tellyzbad1]

There are on average 35,000 frames per episode and you want to manually correct them frame by frame? Are you serious?

He probably thinks film flickering requires manual balancing for it to be a good color correction :/

no some part in the same reels have really differrent colors (even color cast is differrent) you can do a full part perfectly for those episodes with the same settings?

A full episode won't work with one setting, of course...but...i'm not going to need to do it frame by frame

[lansing]

Okay I have uploaded a before/after color matching comparison video here:

https://youtu.be/U82KSgtbc9o

Pay attention to the yellow skins and green, they were impossible to color correct with other tools, and the program did an amazing job matching them.

[Robo4900]

I think CCing to the BDs is pointless.

Even if Steve Franko did indeed do his scan up to that point, and Funi did in fact use that for their BDs(I'm still convinced it's an upscale of some sort), it's pretty clear throughout the BDs that the colours have been tampered with in some way(In general, they're far too blue), and are basically screwed.

[lansing]

I think CCing to the BDs is pointless.

Even if Steve Franko did indeed do his scan up to that point, and Funi did in fact use that for their BDs(I'm still convinced it's an upscale of some sort)

What else can they use? I mean this is simple logical reasoning. You compare the color of the season to all their previous releases, orange bricks, singles, and you can't find anything that look similar, then where do you think this season color come from? It has to be new.

it's pretty clear throughout the BDs that the colours have been tampered with in some way(In general, they're far too blue), and are basically screwed.

I think that's just you. Well it's very easy to check, find an episode that you think has a blue "cast" on it, then find a spot in the episode where you think should be white, like eyes, and then get an eyedropper and sample that spot. If the reading shows a consistent significantly higher blue value, then there's a cast, if not, no cast.

[tellyzbad1]

Robo might be onto something. They may have used the scans they did for the orange bricks. Though they originally claimed the orange bricks had a 1080p scan before being downscaled for DVD, maybe FUNi no longer kept the 1080p data of the scans and just re-upscaled the downscaled scans (and i'm saying they'd kept the downscaled data because much easier for digital storage). A lot of pure assumption going into this but I have my reasons. The "cleanup's" and "enhancement's" intensity is just too high. It wasn't this extreme even on the orange bricks...possibly because orange bricks weren't an upscale project, but these season blurays are? Also, just look at FUNimation's bluray movies. Yes, I know they were 35mm, but the difference between 35mm and 16mm doesn't warrant FUNimation to apply all those filters they did. But an upscale might make them think it's needed.

but yeah who knows, maybe it was the same HD scans as level sets

[lansing]

Robo might be onto something. They may have used the scans they did for the orange bricks. Though they originally claimed the orange bricks had a 1080p scan before being downscaled for DVD, maybe FUNi no longer kept the 1080p data of the scans and just re-upscaled the downscaled scans (and i'm saying they'd kept the downscaled data because much easier for digital storage). A lot of pure assumption going into this but I have my reasons. The "cleanup's" and "enhancement's" intensity is just too high. It wasn't this extreme even on the orange bricks...possibly because orange bricks weren't an upscale project, but these season blurays are? Also, just look at FUNimation's bluray movies. Yes, I know they were 35mm, but the difference between 35mm and 16mm doesn't warrant FUNimation to apply all those filters they did. But an upscale might make them think it's needed.

but yeah who knows, maybe it was the same HD scans as level sets

A Japanese guy invented a way to check it using FFT many years ago, basically he takes in a regular screenshot(no flash, no special effect) from the video, turn it into a FFT image that look something like this:


If the video is true HD, the ball in the middle would be big and bright. If the video is an upscale, the ball would be smaller and dim. He's the one that find out that Fullmetal Alchemist Brotherhood blu-ray was actually an upscale from 960x540, The accuracy of his statement is unknown, but that ugly aliasing throughout the entire FA Brotherhood complemented what he said that the anime was an upscale.

http://anibin.blogspot.com/2009/04/full ... ist-1.html


Another thought to check it is to look for some scenes in the season set that have many details in a very tight space, and then compare that to the same frame in the orange brick. If the frame on the orange brick is blurrier and you couldn't really see the details, then there's no way it's an upscale.

[clutchins]

The Level Sets aren't even the best in terms of color. Blacks are way too crushed and it gives it an overall darker look. If you're going to color correct the show, you need multiple sources from which to cull color information e.g. original broadcasts, Kai, the old single DVDs, and of course the Dragon Box.

[HakkaiBills93]

He probably thinks film flickering requires manual balancing for it to be a good color correction :/

no some part in the same reels have really differrent colors (even color cast is differrent) you can do a full part perfectly for those episodes with the same settings?

A full episode won't work with one setting, of course...but...i'm not going to need to do it frame by frame

so we are okay, it's really small part time to time but it's time consumming doing check everything
the best could have been having a plugin or function that made a automatic frame by frame process a little like acobw but it's impossible

[lansing]

The Level Sets aren't even the best in terms of color. Blacks are way too crushed...

Yes the Level set has crushed black...but not Kai. Kai should have more details that would be visible in those darker areas after the color match compares to the Level set.

and it gives it an overall darker look

The Level set is just following the standard rule of level range used for the anaolog TV, since the anime was made during that era. In analog TV, they don't have the full range we have today, their black was not true black and their white was not 100% white, so it's going to end up with a darker and less contract picture overal compares to one we watch today. But there's nothing wrong about this approach, it looks good on Dragon Ball and I'm going to follow this principle.

If you're going to color correct the show, you need multiple sources from which to cull color information e.g. original broadcasts, Kai, the old single DVDs, and of course the Dragon Box.

I went into a debate on this a few months back in here and I'm going to say it again. This whole idea of taking a piece of this and taking a piece of that, mix them up and call it a product with accurate color is impractical and does not make any kind of sense. How can you make a color accurate reference when none of your source is color accurate? Where is the common sense?

[tellyzbad1]

lol it shouldn't require immense research to know that FMA BH was an upscale? I don't know, maybe it's just me, but with digitally animated animes I can usually tell by eye whether it's been upscaled. A lot of animes from that era which were directly released on bluray looked like upscales to me, e.g. samurai champloo, baccano, gunslinger girl and more... Weirdly, though, when I watch the new ongoing anime Boku no Hero Academia, I deduce a similar "upscaed" look despite me being conscious of the fact that there's no need for it to be an upscale...lol...

going back to dragon ball colors, i'm surprised as to why there's still so much debate on what colors to use? With Dragon Boxes, you will never be able to have as much dark details as there should be, proven by the huge amount of detail revealed in Kai. So, at the end of the day, you're not going to ever have a great product through cc'ing Dbox so no point bogging down on trying to make it extremely definitive. sometimes it's like you guys want to find the exact hex-codes of the colors used in every single pixel from whatever it is you're regarding as the "original colors" lol. when really all u have to do is remove color cast and then reduce the filmic presence from the colors.

this "filmic presence" can be taken out by simply making the cyans more magenta and the reds more yellow. sadly, though, turning reds into yellow won't give you the ideal result. it fixes one issue, but causes another. it'll fix goku's gi, but cause actual reds (like jeice's skin, or gohan's red belt) to appear orange. so what we need is a tool that can precisely distinguish red-oranges from actual reds and add the yellow only to the red-oranges. Q-tec could do this because the raw film offers them to work with higher color space while we only have the 8-bit video but yeah...upon doing those things, you've pretty much fixed the DBox colors. Only other thing i can think of is goku's blue top looking a bit too purple and less aesthetic than the purer blue seen in Kai and some cels...but, i've seen the purple-ish tint of the blue top inn non-film based shots such as toei vod thumbnails

[lansing]

going back to dragon ball colors, i'm surprised as to why there's still so much debate on what colors to use? With Dragon Boxes, you will never be able to have as much dark details as there should be, proven by the huge amount of detail revealed in Kai. So, at the end of the day, you're not going to ever have a great product through cc'ing Dbox so no point bogging down on trying to make it extremely definitive. sometimes it's like you guys want to find the exact hex-codes of the colors used in every single pixel from whatever it is you're regarding as the "original colors" lol. when really all u have to do is remove color cast and then reduce the filmic presence from the colors.

Beside from the fact that this idea does not make sense, additionally, the way I see it is that they are not practical. When color correcting something with magnitude like dragon ball, you cannot do it alone, you have to do it as a community. And in order to do that, the project has to be scalable and be able to break down into smaller pieces so that everybody can contribute. And to make it doable for everyone, each job has to be easy.

Now, how on earth are you going to convert that "color accurate from mixed sources" idea into a project like this? Let's say you DO figure out a way to make the idea happen, then what's next? How do you scale that up? Well you can't split it up because you are the only one that know how to do it. So are you going to teach 500 people the whole process and then have them repeat the process frame by frame? This is what I call impractical, only theories and talks, and no practical plan for execution.

[HakkaiBills93]

lol it shouldn't require immense research to know that FMA BH was an upscale? I don't know, maybe it's just me, but with digitally animated animes I can usually tell by eye whether it's been upscaled. A lot of animes from that era which were directly released on bluray looked like upscales to me, e.g. samurai champloo, baccano, gunslinger girl and more... Weirdly, though, when I watch the new ongoing anime Boku no Hero Academia, I deduce a similar "upscaed" look despite me being conscious of the fact that there's no need for it to be an upscale...lol...

going back to dragon ball colors, i'm surprised as to why there's still so much debate on what colors to use? With Dragon Boxes, you will never be able to have as much dark details as there should be, proven by the huge amount of detail revealed in Kai. So, at the end of the day, you're not going to ever have a great product through cc'ing Dbox so no point bogging down on trying to make it extremely definitive. sometimes it's like you guys want to find the exact hex-codes of the colors used in every single pixel from whatever it is you're regarding as the "original colors" lol. when really all u have to do is remove color cast and then reduce the filmic presence from the colors.

this "filmic presence" can be taken out by simply making the cyans more magenta and the reds more yellow. sadly, though, turning reds into yellow won't give you the ideal result. it fixes one issue, but causes another. it'll fix goku's gi, but cause actual reds (like jeice's skin, or gohan's red belt) to appear orange. so what we need is a tool that can precisely distinguish red-oranges from actual reds and add the yellow only to the red-oranges. Q-tec could do this because the raw film offers them to work with higher color space while we only have the 8-bit video but yeah...upon doing those things, you've pretty much fixed the DBox colors. Only other thing i can think of is goku's blue top looking a bit too purple and less aesthetic than the purer blue seen in Kai and some cels...but, i've seen the purple-ish tint of the blue top inn non-film based shots such as toei vod thumbnails

that's why i didn't use avisynth i could have used it if i could use the same filter as the after effect selective color as this one give you more freedom with colors

[tellyzbad1]

going back to dragon ball colors, i'm surprised as to why there's still so much debate on what colors to use? With Dragon Boxes, you will never be able to have as much dark details as there should be, proven by the huge amount of detail revealed in Kai. So, at the end of the day, you're not going to ever have a great product through cc'ing Dbox so no point bogging down on trying to make it extremely definitive. sometimes it's like you guys want to find the exact hex-codes of the colors used in every single pixel from whatever it is you're regarding as the "original colors" lol. when really all u have to do is remove color cast and then reduce the filmic presence from the colors.

Beside from the fact that this idea does not make sense, additionally, the way I see it is that they are not practical. When color correcting something with magnitude like dragon ball, you cannot do it alone, you have to do it as a community. And in order to do that, the project has to be scalable and be able to break down into smaller pieces so that everybody can contribute. And to make it doable for everyone, each job has to be easy.

Now, how on earth are you going to convert that "color accurate from mixed sources" idea into a project like this? Let's say you DO figure out a way to make the idea happen, then what's next? How do you scale that up? Well you can't split it up because you are the only one that know how to do it. So are you going to teach 500 people the whole process and then have them repeat the process frame by frame? This is what I call impractical, only theories and talks, and no practical plan for execution.

What on earth are you talking about? How does what I'm saying not make sense? Everything you've said there has nothing to do with what I said?

The purpose of what I said wasn't to address what you guys can do as a community? I'm addressing this quandary on what the final product should look like... You guys can't seem to figure out whether you want to use the season blurays, the level sets or clutchins' "you have to use multiple sources"... I'm telling you what's wrong witht he DBox colors and how to remove those wrongs in the simplest way possible without having to hunt down some ideal, non-existent reference source (and this simplifying actually helps this "community project" thing you're trying to achieve).

I don't get how you say "it's not practical". Legit just remove color cast and then increase magenta in cyan and increase yellow in reds...I've literally told you exactly what you have to do? Nowhere do I mention what I say involving frame by frame work or whether it has to be done by one person or a group. Though, me personally, if I desire a high-quality product, I am more in favor of letting one competent person create it, rather than a thousand joe schmoes.

[lansing]

going back to dragon ball colors, i'm surprised as to why there's still so much debate on what colors to use? With Dragon Boxes, you will never be able to have as much dark details as there should be, proven by the huge amount of detail revealed in Kai. So, at the end of the day, you're not going to ever have a great product through cc'ing Dbox so no point bogging down on trying to make it extremely definitive. sometimes it's like you guys want to find the exact hex-codes of the colors used in every single pixel from whatever it is you're regarding as the "original colors" lol. when really all u have to do is remove color cast and then reduce the filmic presence from the colors.

Beside from the fact that this idea does not make sense, additionally, the way I see it is that they are not practical. When color correcting something with magnitude like dragon ball, you cannot do it alone, you have to do it as a community. And in order to do that, the project has to be scalable and be able to break down into smaller pieces so that everybody can contribute. And to make it doable for everyone, each job has to be easy.

Now, how on earth are you going to convert that "color accurate from mixed sources" idea into a project like this? Let's say you DO figure out a way to make the idea happen, then what's next? How do you scale that up? Well you can't split it up because you are the only one that know how to do it. So are you going to teach 500 people the whole process and then have them repeat the process frame by frame? This is what I call impractical, only theories and talks, and no practical plan for execution.

What on earth are you talking about? How does what I'm saying not make sense? Everything you've said there has nothing to do with what I said?

The purpose of what I said wasn't to address what you guys can do as a community? I'm addressing this quandary on what the final product should look like... You guys can't seem to figure out whether you want to use the season blurays, the level sets or clutchins' "you have to use multiple sources"... I'm telling you what's wrong witht he DBox colors and how to remove those wrongs in the simplest way possible without having to hunt down some ideal, non-existent reference source (and this simplifying actually helps this "community project" thing you're trying to achieve).

I don't get how you say "it's not practical". Legit just remove color cast and then increase magenta in cyan and increase yellow in reds...I've literally told you exactly what you have to do? Nowhere do I mention what I say involving frame by frame work or whether it has to be done by one person or a group. Though, me personally, if I desire a high-quality product, I am more in favor of letting one competent person create it, rather than a thousand joe schmoes.

I wasn't referring to you, I was referring to the people who said they can make a color accurate reference from a mix of sources. I was complementing on what you said by adding reasons to why their way does not work.

[ect5150]

... and Funi did in fact use that for their BDs(I'm still convinced it's an upscale of some sort)

An upscale of sorts? An upscale of what? the Orange Bricks? The DBoxes? There is more detail in the BluRays than even in some of the Dragon Box episodes. THis is especially apparent when you look at the old Level Set releases.

[tellyzbad1]

... and Funi did in fact use that for their BDs(I'm still convinced it's an upscale of some sort)

An upscale of sorts? An upscale of what? the Orange Bricks? The DBoxes? There is more detail in the BluRays than even in some of the Dragon Box episodes. THis is especially apparent when you look at the old Level Set releases.

i think if it was an upscale of some sort, we could assume that when they originally scanned the film in 1080p for orange bricks, they then downscaled the film scans before the cleanup processes and transferring to DVD. then, they got rid of the digital files for the 1080p scans but kept the digital files for the 480p scans because they're easier to store on hard drives or whatever. and these season blurays maybe upscaled and filtered versions of the downscaled scans and filtered them heavily.

It's pure assumption that they would downscale the scans and kept them in digital storage, but I want to believe that all the filtering and sharpening done in the season blurays was to fake a HD look.