Otis Grapsas
Carbonite Member
www.dxomark.com have some good dynamic range measurement in still mode.
Discuss Cinematography in this Awesome Rihanna Video
- Thread starter Homunculus
- Start date
GRENCH
Well-known member
Don't blame me. I'm just the messenger
Don't blame me. I'm just the messenger
I'm not telling you anything. You saw the footage for yourself. Like I said I'm just the messenger. I didn't conduct the test personally.
I guess what it comes down to is who do you trust. The shootout was the only test that I had the fortune to see and to me it looked as if it was done correctly. I doubt if these guys would put their reputations on the line for a latitude test.
Don't blame me. I'm just the messenger
I'm not telling you anything. You saw the footage for yourself. Like I said I'm just the messenger. I didn't conduct the test personally.
I guess what it comes down to is who do you trust. The shootout was the only test that I had the fortune to see and to me it looked as if it was done correctly. I doubt if these guys would put their reputations on the line for a latitude test.
Last edited:
Patryk_Rebisz
Veteran
If anything that Zacuto shootout test proves how superior film is to digital.
And comparing latitude of DSLR in still mode and video mode - those are two tally different things, just because the sensor can resolve X amount of stops (still mode) doesn't mean it translates to highly compressed video mode (it does not).
And comparing latitude of DSLR in still mode and video mode - those are two tally different things, just because the sensor can resolve X amount of stops (still mode) doesn't mean it translates to highly compressed video mode (it does not).
GRENCH
Well-known member
The DSLRs were tested in Video mode from what I could see. None of the test were conducted in still mode. You could clearly see hat they were not still images. Please correct me if I'm wrong. And based on the conversations that were shown in these test I believe it showed the pros and cons of both. If you watched all three episodes you would have seen how the Canon 1D and the Nikon D3 blew film out of the water in the foot candle light test (a man holding a light from 2 foot in front of him and then moving it closer).
Otis Grapsas
Carbonite Member
The dSLR video mode is sometimes crushed in the lower stops vs the still mode but the real problem is compression is so bad the lower stops end up with a shade that constantly changes + compression artifacts. A stop is only useful photographically if it has texure and detail. A video camera with the same dynamic range will be more useful because it will be able to take some post processing without breaking apart. It does not have to be uncompressed or HDCAM-SR. Even AVCHD is better than what dSLRs have to offer. The Zacuto test ignores that parameter.
The dynamic range part of the test is laughable. The video presented does not even agree with their comments and results. And they ignore how bad the compression is. Even in the compressed low resolution version they present on the web, film is far superior due to grain vs compression artifacts.
Film is scanned and ends up as an intraframe format, sometimes even uncompressed. Incredible how the Zacuto tests ignore motion quality. They also ignore how better film shutter is to the dSLR rolling shutter and compression. Film will give you 1/96sec if you ask for it, a dSLR will give you something that looks like 1/24sec due to the compression and filtering. Just like HDV cannot provide the true shutter look, only worse. Basically, that Zacuto test was designed to prove that dSLRs are high quality imaging devices that cinematographers are excited about. Zacuto need that in order to sell dSLR accessories.
The dynamic range part of the test is laughable. The video presented does not even agree with their comments and results. And they ignore how bad the compression is. Even in the compressed low resolution version they present on the web, film is far superior due to grain vs compression artifacts.
Film is scanned and ends up as an intraframe format, sometimes even uncompressed. Incredible how the Zacuto tests ignore motion quality. They also ignore how better film shutter is to the dSLR rolling shutter and compression. Film will give you 1/96sec if you ask for it, a dSLR will give you something that looks like 1/24sec due to the compression and filtering. Just like HDV cannot provide the true shutter look, only worse. Basically, that Zacuto test was designed to prove that dSLRs are high quality imaging devices that cinematographers are excited about. Zacuto need that in order to sell dSLR accessories.
Patryk_Rebisz
Veteran
GRENCH, are you watching the same video i'm watching as they didn't even test film in that specific set-up.
Look at the shots with the girl in the bathtub shot on film - you can see details in the window behind (not with DSLRs). Look at the shot of the guy reading the book - you can see details of the light bulb with film (not with DSLRs). IF it doesn't prove to you that film wins when latitude is concerned - i don't know what will.
Look at the shots with the girl in the bathtub shot on film - you can see details in the window behind (not with DSLRs). Look at the shot of the guy reading the book - you can see details of the light bulb with film (not with DSLRs). IF it doesn't prove to you that film wins when latitude is concerned - i don't know what will.
PerroneFord
Deals in Lead
Patryk,
Film wins the latitude battle, pure and simple... all day every day. Where the DSLRs and other large sensor digital cameras win is in other areas. Like when you need to be clean at 800ASA and beyond. Or if you need to try to bring something out of the shadows. Film does a terrific job at holding highlights, but it doesn't do nearly so well in the shadows or the foot. The situation is essentially reversed for digital. Thus, you really have to look at your production and make the choice.
If cost, is an issue, then film is nearly a non-starter. It's prohibitely expensive for most people.
All that aside, whether the DSLRs are 7 stops, 8 stops, or 11 stops. The fact that they can even be shown seriously on the same screen as modern film is testament to the images they are making below $3k, and in some cases below $1k. Compared to what we had a year ago this time at those prices, it's utterly remarkable. Yes, there are lots better solutions, but the nearest, ready to shoot camera that's close is $50k or so. That's a pretty darn big leap for a small producer.
Film wins the latitude battle, pure and simple... all day every day. Where the DSLRs and other large sensor digital cameras win is in other areas. Like when you need to be clean at 800ASA and beyond. Or if you need to try to bring something out of the shadows. Film does a terrific job at holding highlights, but it doesn't do nearly so well in the shadows or the foot. The situation is essentially reversed for digital. Thus, you really have to look at your production and make the choice.
If cost, is an issue, then film is nearly a non-starter. It's prohibitely expensive for most people.
All that aside, whether the DSLRs are 7 stops, 8 stops, or 11 stops. The fact that they can even be shown seriously on the same screen as modern film is testament to the images they are making below $3k, and in some cases below $1k. Compared to what we had a year ago this time at those prices, it's utterly remarkable. Yes, there are lots better solutions, but the nearest, ready to shoot camera that's close is $50k or so. That's a pretty darn big leap for a small producer.
GRENCH
Well-known member
?
?
I know. You just proved my point. Based on the test and that test alone there was no point in testing film because film would not work in that foot candle test. These are the words they spoke. They stated that testing film in that test would be pointless.
I never said DSLR's were superior. I simply stated that in the Zacuto shoot out they stated that the Canon 7D hit 11 stops. Like I said earlier. I'm just the messenger.
?
I know. You just proved my point. Based on the test and that test alone there was no point in testing film because film would not work in that foot candle test. These are the words they spoke. They stated that testing film in that test would be pointless.
I never said DSLR's were superior. I simply stated that in the Zacuto shoot out they stated that the Canon 7D hit 11 stops. Like I said earlier. I'm just the messenger.
Ryan Patrick O'Hara
Veteran
I know. You just proved my point. Based on the test and that test alone there was no point in testing film because film would not work in that foot candle test. These are the words they spoke. They stated that testing film in that test would be pointless.
You are right. It was really nice of Canon to give John Alcott and Stanley Kubrick a very early prototype of a Canon Video-DSLR for Barry Lyndon. I mean, how could they have ever shot those scenes on celluloid film?!
Last edited:
Homunculus
Carbonite Member
Grench I know you're just saying what you saw in the zacuto shootout. All I'm saying is that there's something wrong with either their results or whatever because I'm pretty certain a clean 11stops is pretty much impossible for DSLR considering that's Red One territory and even my untrained eye can clearly see in side by side footage that DSLR's don't even come close to Red One dynamic range. The whole point of the $5,000+ MX sensor upgrade for the Red One was to get its DR bumped from 11 some odd to 13+ stops.
also should be stated that a DR test is very arbitrary and subjective. Reason being that everyone has a different view on where the line ends and what becomes usable or too noisy. Even Red admitted on reduser.net that MX sensor can be seen by some to go as high as 14 or 14.5 but others would conservatively put it at 13 because at 14+ stops it is too noisy for some at its limits.
So who knows, perhaps 7D had SOME resolution nearing the 11 stop range? Perhaps the conditions in which they shot were somehow idealized or fixed, I have no idea but I agree with the others that Zacuto is after all a DSLR 'parts dealer' and as such it is in their best interest to make DSLR's look as good as possible, thus hiring guys like Shane Halbut and PHilip Bloom amongst others to represent the show.
also should be stated that a DR test is very arbitrary and subjective. Reason being that everyone has a different view on where the line ends and what becomes usable or too noisy. Even Red admitted on reduser.net that MX sensor can be seen by some to go as high as 14 or 14.5 but others would conservatively put it at 13 because at 14+ stops it is too noisy for some at its limits.
So who knows, perhaps 7D had SOME resolution nearing the 11 stop range? Perhaps the conditions in which they shot were somehow idealized or fixed, I have no idea but I agree with the others that Zacuto is after all a DSLR 'parts dealer' and as such it is in their best interest to make DSLR's look as good as possible, thus hiring guys like Shane Halbut and PHilip Bloom amongst others to represent the show.
Homunculus
Carbonite Member
by the way speaking of latitude, is it possible to get increased latitude from a lens? I would think this is impossible but why does Shane Hurlbut say that he got 2 extra stops of latitude from switching lenses on his Canon 5dmkii? http://provideocoalition.com/index.php/cmilby/story/shane_hurlbut/P1/
Ryan Patrick O'Hara
Veteran
I think that has more to do with the contrast of an image, which can effect many things, but not technically the DR of a camera sensor... however it looks like more. Which then begins a chicken and egg argument where you must ask yourself, in cinematography, if something looks to do something, then what's the difference.
No pun intended, but you get into muddy water when you start thinking like that. Would putting a contrast filter in front of the lens to get light to bleed into the shadows count as improved DR/latitude? You'd gain more information in the shadows, but you'd lose the deep blacks, which DOES account for part of the range, right? So it would rememble to have more mid-low range, but you'd be loosing the really low areas. It's more of a contrast issue. Right?
I hope that makes sense.
No pun intended, but you get into muddy water when you start thinking like that. Would putting a contrast filter in front of the lens to get light to bleed into the shadows count as improved DR/latitude? You'd gain more information in the shadows, but you'd lose the deep blacks, which DOES account for part of the range, right? So it would rememble to have more mid-low range, but you'd be loosing the really low areas. It's more of a contrast issue. Right?
I hope that makes sense.
Homunculus
Carbonite Member
ah I see what you're saying.
Hurlbut says a few very interesting things in that interview. Made my eyes boggle wide open at first but upon review seems to sort of make sense.
For instance he says that he absolutely LOVES the compression of DSLR's (or rather the canon 5d in particular) because it makes skin tones look better and more natural whereas in non compressed HD footage he says skin tones look plastic, or something along those lines.
Wonder what peoples' thoughts on this are? This is the first person I've ever in my life heard say that they like compression, and it's an ASC top cinematographer at that.
Is Canon paying this guy something nasty to be saying such things or what?
Hurlbut says a few very interesting things in that interview. Made my eyes boggle wide open at first but upon review seems to sort of make sense.
For instance he says that he absolutely LOVES the compression of DSLR's (or rather the canon 5d in particular) because it makes skin tones look better and more natural whereas in non compressed HD footage he says skin tones look plastic, or something along those lines.
Wonder what peoples' thoughts on this are? This is the first person I've ever in my life heard say that they like compression, and it's an ASC top cinematographer at that.
Is Canon paying this guy something nasty to be saying such things or what?
PerroneFord
Deals in Lead
I've seen ASC guys say a lot of things. And they weren't being paid by anyone to say it. They're just like us. Some like things a certain way, others like it a different way. I remember reading about the DP on the recent movie 21 saying he rated the Genesis at ISO 320, whereas the DP for Apocolypto rated the same camera at ISO 500! They don't even agree on camera basics, much less aesthetic.
Some DPs love the look of the DI. Others hate it and won't take a movie through it. Some prefer the look of film, others prefer the clean look of digital. So it doesn't surprise me one bit if Hurlbut says he likes the compression. He prefers a LOT more edgy picture than many of his peers. That's why you find him on the pictures you do. Things like Terminator Salvation. I probably wouldn't hire him for a period piece...
The two movies I've DPed on the 5D have both been strongly character driven, so I've had faces in the camera for most of both shoots. I don't see anything particularly fabulous there, but the tones are somewhat pleasing. Honestly, I think a LOT has to do with the glass. I this is something I think people REALLY overlook when we discuss the look of these cameras. The glass you put in front of that sensor, including any filters, is REALLY what is shaping that image. If the sensor does it's job well, it's merely the canvas on which you are painting. The lens is the brush, and the light is the paint itself. Some artists prefer to do watercolor on paper, others prefer to do oils on canvas. Some prefer the organic nature of film, others don't. But the glass shapes and bends every beam of light that hits that sensor and simply cannot be overlooked.
One thing I learned VERY quickly reading American Cinematographer, was just how carefully the DPs selected glass. More so that lighting even. Some go so far as to rehouse lenses for use on different cameras. I remember reading how the DP on "There Will be Blood" rehoused some lenses and sought out very specific lenses that were VERY old. Like 50-70 years old. Yes, the film imparted a look, but they spent months getting the lenses they wanted. These guys will hold up a production based on lens availability. Think about that the next time you're going cheap on your glass for your work.
Some DPs love the look of the DI. Others hate it and won't take a movie through it. Some prefer the look of film, others prefer the clean look of digital. So it doesn't surprise me one bit if Hurlbut says he likes the compression. He prefers a LOT more edgy picture than many of his peers. That's why you find him on the pictures you do. Things like Terminator Salvation. I probably wouldn't hire him for a period piece...
The two movies I've DPed on the 5D have both been strongly character driven, so I've had faces in the camera for most of both shoots. I don't see anything particularly fabulous there, but the tones are somewhat pleasing. Honestly, I think a LOT has to do with the glass. I this is something I think people REALLY overlook when we discuss the look of these cameras. The glass you put in front of that sensor, including any filters, is REALLY what is shaping that image. If the sensor does it's job well, it's merely the canvas on which you are painting. The lens is the brush, and the light is the paint itself. Some artists prefer to do watercolor on paper, others prefer to do oils on canvas. Some prefer the organic nature of film, others don't. But the glass shapes and bends every beam of light that hits that sensor and simply cannot be overlooked.
One thing I learned VERY quickly reading American Cinematographer, was just how carefully the DPs selected glass. More so that lighting even. Some go so far as to rehouse lenses for use on different cameras. I remember reading how the DP on "There Will be Blood" rehoused some lenses and sought out very specific lenses that were VERY old. Like 50-70 years old. Yes, the film imparted a look, but they spent months getting the lenses they wanted. These guys will hold up a production based on lens availability. Think about that the next time you're going cheap on your glass for your work.
Otis Grapsas
Carbonite Member
The MTF does affect dynamic range measurements, the lens compresses the image contrast. An 100% MTF does nothing to the dynamic range at a given frequency, a 50% increases it by 1 stop, a 25% by two, a 12.5% by three etc.
A typical lens will be very high in MTF at low frequencies though so the effect will be limited to high frequencies and possibly corners in wide apertures.
In order to get two extra stops from a 5dII by using the low contrast of the lens, you will need a lens that is no more than 25% MTF at ALL frequencies. Even 0.1lp/mm. The 5dII video mode operates at a resolution of 26.6lp/mm. You can safely assume that any modern lens is not only high in MTF at 0.1lp/mm, it's also very high in MTF up to 30 lp/mm, so there in no advantage whatsoever. I doubt the cine lenses that article mentions have low MTF at low frequencies.
You can see a test using a cheap lens here:
http://photo.net/equipment/canon/efs18-55/
The effect of the low MTF is hardly noticeable even though the MTF is down to 25% in the corners at 30lp/mm. The sky is a little darker in soft apertures, but nothing that important. That's because the lower frequencies have a much higher MTF compared to the 10 and 30 line curves shown in the graphs. Think of a flat 99% MTF at 1lp/mm and lower, and 1lp/mm is still smaller than the overexposed window in your shot! And that's a cheap zoom at the full resolution of the sensor in still mode, not in video mode. In video mode the effective pixel size is much larger and the MTF vs resolution requirements far lower so MTF is very high.
A typical lens will be very high in MTF at low frequencies though so the effect will be limited to high frequencies and possibly corners in wide apertures.
In order to get two extra stops from a 5dII by using the low contrast of the lens, you will need a lens that is no more than 25% MTF at ALL frequencies. Even 0.1lp/mm. The 5dII video mode operates at a resolution of 26.6lp/mm. You can safely assume that any modern lens is not only high in MTF at 0.1lp/mm, it's also very high in MTF up to 30 lp/mm, so there in no advantage whatsoever. I doubt the cine lenses that article mentions have low MTF at low frequencies.
You can see a test using a cheap lens here:
http://photo.net/equipment/canon/efs18-55/
The effect of the low MTF is hardly noticeable even though the MTF is down to 25% in the corners at 30lp/mm. The sky is a little darker in soft apertures, but nothing that important. That's because the lower frequencies have a much higher MTF compared to the 10 and 30 line curves shown in the graphs. Think of a flat 99% MTF at 1lp/mm and lower, and 1lp/mm is still smaller than the overexposed window in your shot! And that's a cheap zoom at the full resolution of the sensor in still mode, not in video mode. In video mode the effective pixel size is much larger and the MTF vs resolution requirements far lower so MTF is very high.
Otis Grapsas
Carbonite Member
Did he actually say that? It's the other way around. Compression makes things look like plastic because it cannot maintain texture. The chroma space in poor compression quantises the color space and reduces tonality. The alias in a dSLR does not help texture either.
"Legendary ASC discusses how the HDSLR changed his life, and how it will change yours…" :laugh:
GRENCH
Well-known member
Please don't shoot!!! I have a message.
Please don't shoot!!! I have a message.
This is more like it. Now we are holding a conversation:cheesy:
Thank you for not shooting the messenger
I'm not saying that the Zucuto test is the bible of test. It's just the only one of this type that I have seen. I have edited RED footage and I do agree that RED has far more dynamic range than DSLR's. But RED was not in that shoot out (maybe for obvious reasons). I simply quoted what was said about those cameras that were in the Zacuto shoot out. Whether the 7D's 11 stop point has too much noise in it to be usable is another discussion entirely. I never stated that at 11 stops the 7D's image would be worthy of praise.
I'm a firm believer in truth or no truth. A half a truth is still a lie.
Hilarious All I can do is quote myself here "Based on the test and that test alone there was no point in testing film because film would not work in that foot candle test. These are the words they spoke. They stated that testing film in that test would be pointless."
Now if they are wrong then forgive me for co signing that as fact but with no light and just a lighter I was under the impression that it would be impossible to pull that off.
Please don't shoot!!! I have a message.
This is more like it. Now we are holding a conversation:cheesy:
Thank you for not shooting the messenger
I'm not saying that the Zucuto test is the bible of test. It's just the only one of this type that I have seen. I have edited RED footage and I do agree that RED has far more dynamic range than DSLR's. But RED was not in that shoot out (maybe for obvious reasons). I simply quoted what was said about those cameras that were in the Zacuto shoot out. Whether the 7D's 11 stop point has too much noise in it to be usable is another discussion entirely. I never stated that at 11 stops the 7D's image would be worthy of praise.
Now if you guys are saying that these guys are benders the truth (liars) then that is another argument entirely that I am not willing to be apart of due to my own lack of knowledge in some of these areas.
I'm a firm believer in truth or no truth. A half a truth is still a lie.
Hilarious
All I can do is quote myself here "Based on the test and that test alone there was no point in testing film because film would not work in that foot candle test. These are the words they spoke. They stated that testing film in that test would be pointless."Now if they are wrong then forgive me for co signing that as fact but with no light and just a lighter I was under the impression that it would be impossible to pull that off.
Last edited:
Homunculus
Carbonite Member
Yes hard to pull off on film with typical t2.0 or t1.4 lenses but Barry Lyndon was shot with legendary lens made by Nasa as a gift to Kubrick that is T/0.7 so that is quite an order of magnitude higher aperture than what anything else is capable of be it DSLR, film, whatever
PerroneFord
Deals in Lead
Um, T0.7 is 2 stops faster than the T1.4 they had available. You can buy glass today off the shelf that is within a stop of that.
This is not even a single order of magnitude. It's merely 4x faster than T1.4