Kinefinity Terra 4K

[Cary Knoop]

Originally Posted by Tom Roper

I don't want to seem unfair, so I'll segue my previous remark and stop there. But if you look around the 18 second mark of the Luna Kids video where the lady is paying for her crepe, you can see both characteristics at once, unremarkable DR and highlight treatment but also something else that I think is contributing to the overall look; I don't see pure black anywhere! Not in that scene, not in any of the videos!

I think these are all a matter of grading. I think pure black is just as bad, especially if the roll-off is steep, as pure white unless you have really pure black or white elements in the scene.
Looking at the 18 seconds in, if you check the scopes you see that the person who color corrected this video is leaving super whites. Pulling them down to legal does actually remove some (but not all) of the clipping. Looking at the scopes blacks seem to have been raised in post.

So I am not particularly convinced this was the best grading.

Left picture is a detail of the original file, the right picture is the same detail but after the lowering of the super-whites and lowering the blacks.

[Gary Huff]

Originally Posted by Cary Knoop

Dynamic range and the colors. Also, it grades very well.

What do you mean by the dynamic range specifically? How about the colors? What about the colors...specifically?

What do you mean by "grades"? What are you doing to the Kinefinity ProRes (or cDNG) clips exactly? What do other formats (such as the H.264-based codecs from other cameras) do specifically when you try to do the same kind of process as you did with the Kinefinity clips?

[Gary Huff]

Originally Posted by Cary Knoop

Pulling them down to legal does actually remove some of the clipping. Looking at the scopes blacks seem to have been raised in post.

How would you know this based on what is, most likely, a second-generation 8-bit 4:2:0 H.264 encode? I would not trust ripping a clip from YouTube to tell me anything about the original source.

[Cary Knoop]

Originally Posted by Gary Huff

What do you mean by the dynamic range specifically? How about the colors? What about the colors...specifically?

What do you mean by "grades"? What are you doing to the Kinefinity ProRes (or cDNG) clips exactly? What do other formats (such as the H.264-based codecs from other cameras) do specifically when you try to do the same kind of process as you did with the Kinefinity clips?

You know I am not going to be in the defense seat, it's clear you have another opinion.
I have a feeling you do not respect my opinion at all, so what is the point?

I gave my opinion, I like this camera very much!
I guess I must be guilty as charged!

Originally Posted by Gary Huff

How would you know this based on what is, most likely, a second-generation 8-bit 4:2:0 H.264 encode? I would not trust ripping a clip from YouTube to tell me anything about the original source.

I base it on the encode the upload contains super-whites. Simple as that! Who, except you, is talking about the original source!

[Gary Huff]

Originally Posted by Cary Knoop

You know I am not going to be in the defense seat, it's clear you have another opinion. I have a feeling you do not respect my opinion at all, so what is the point?

I can't ascertain if your opinion is worthwhile because of answers like this. When you say you like the colors and dynamic range and that it grades well, I don't know what you mean. It could all be confirmation bias. And in "grading", I suspect that you mean adjusting contrast and saturation, perhaps every once in a while adding a LUT. That's not really grading, and any camera, even AVCHD, can easily handle those operations.

I gave my opinion, I like this camera very much!

That's fair, you can like whatever you like for whatever reason, but it does not detract from the idea that the Terra 4K doesn't have anything I don't feel you can already get with a GH5 or a Blackmagic Pocket Camera or an A7III.

I base it on the encode the upload contains super-whites. Simple as that! Who, except you, is talking about the original source!

How did you download the video that you tried to fix the grade? 4KDownloader to pull the YouTube encode? Which is probably a re-encode of what is probably a YouTube preset export from whatever the NLE was? How would that be helpful to determine super-whites and whether IR contamination is a problem?

[Tom Roper]

Originally Posted by Gary Huff

How did you download the video that you tried to fix the grade? 4KDownloader to pull the YouTube encode? Which is probably a re-encode of what is probably a YouTube preset export from whatever the NLE was? How would that be helpful to determine super-whites and whether IR contamination is a problem?

I don't know but if I ventured a guess, Cary used a BMD Decklink capture card. He is savvy, his knowledge and grading abilities are without question. So what he is saying is that he got a hold of the source by whatever means, and it contained super whites. It's another way of saying if the source Ycbcr 10 bit word contained data above code value 940 or below 64 (blacker than black), the data points would be clipped from viewing. He made that point by pulling the highlights below 940 where they reappeared.

There is not a conclusion about DR that can be made from this in isolation, just that the video isn't technically graded well. The colorist made the overall image to his liking but in so doing clipped the highlights. You just have to take any observation about DR as an opinion in the absence of hard testing, which itself has some subjectivity and maybe even that has a grain of salt. Remember the hue and cry when Alan Roberts of BBC, the most renowned tester in the world using the latest methods contained in EBU Tech 3335 came up with a DR for the GH5s of 14.6 stops, he was ridiculed, but as far as I know, not disproven.

It doesn't change that I am not as enthusiastic about the Kinefinity , but I might be more inclined by video with properly recorded levels and not the obvious clipped zones I saw in these, and saw some un-hued blacks.

[Gary Huff]

Originally Posted by Tom Roper

I don't know but if I ventured a guess, Cary used a BMD Decklink capture card.

How would using a Decklink to capture a YouTube video add anything to a video that might be playing back as a second generation 8-bit 4:2:0 H.264 encode?

[Tom Roper]

Originally Posted by Gary Huff

How would using a Decklink to capture a YouTube video add anything to a video that might be playing back as a second generation 8-bit 4:2:0 H.264 encode?

It could not add anything per se, except super whites and blacker than black are still applicable to 8 bit, 16-235, even 420 VP9. His position was that highlight detail was revealed in the super white region which would be above 235. While those data points might not have been displayed, they could still have been conveyed in the 8 bit word. That said, I'm speculating because he hasn't told us how he got hold of the source. All we know is that he demonstrated there was clipped data that he recovered. So if that's the basis for his opinion, no one else has the same benefit of it and these particular YouTube videos are worthless for critical evaluation, which well may be the case.

I'm sure this happens a lot, not just YouTube. There are so many places where legal levels can go wrong. And they get "fixed" wrong as this video demonstrates. In other words it can look flat or crushed to the colorist because the levels are wrong, but his fix is to use the grading wheels.

So I don't really know, it's his prerogative to explain or not, but I am certain that he is correct for finding the buried details in the super whites. And I am correct for noticing in several videos now, that DR and highlights were unremarkable. If people don't make critical observations and otherwise just roll along with the excitement and hype, we don't advance the science nor improve the art, or make bad purchase decisions.

[Troy Moss]

Someone recently posted this (strange, but someone might be interested to see 240 fps)........(click on top to go to link)....

ＷＡＶＥＳ on Vimeo

[Cary Knoop]

Originally Posted by Troy Moss

Someone recently posted this (strange, but someone might be interested to see 240 fps)........(click on top to go to link)....

ＷＡＶＥＳ on Vimeo

While we are talking levels, above is an opposite case, the video output is video levels upon video levels.

(Note that the scope in DaVinci Resolve reports 0-1023 as the levels-independent visible range where black is 0 and white is 1023, this range gets properly converted to limited or full range depending on the delivery settings).

[Troy Moss]

80% of this footage is Kinefinity Terra 4K, the other 20% is EVA1 (click on top of link that reads "Punkt on Vimeo").

Thanks Sigurd Neby!

Punkt. on Vimeo