Rodney Charters, ASC - RED ONE, Canon 7D and 5D MkII side by side

[Andrew Walker]

They may store their video in such a dimensioned video, but they measure significantly less... And the codec itself robs more detail from the image.

Graeme

Oh I know, I was just correcting his post. Do you know what the measured resolution is of the 5D2/7D shooting video? Of course to be fair I remember hearing people complain that RED wasn't really 4K and that it was 3.2K. As far as measured resolution goes.

Personally I don't care because you can look at the images coming off of the 5D2/7D vs the RED and see there's a big difference.

 

[Sanjin Jukic]

Canon Canon 1D Mk4, 5D2MK and 7D could be B cameras only

just because of horrible aliasing (moiré), H264 codec issues, also difficulties to use auto focus feature (too slow) and manual focus (not so easy with its small LCD).

That means for example if you don't have a good shot with A cam (lets's say RED1) that you must be really lucky to get a proper shot with those Canon video DSLRs.

Canon 7D aliasing issue (moiré), roof of St. Stephen's Cathedral (German: Stephansdom) in Vienna (enlarged detail).

Canon 7D aliasing issue (moiré), facade of Julius Meinl shop at Graben in Vienna (enlarged detail).

But the footage would tell you even better about this Canon 7D aliasing issue:

Canon 7D aliasing issue (moiré)>>example clip to download @ 1080p H264>>>

 

[Bill Anderson]

The tests show pretty much what I expected: that there
might be a scenario where, every alternative failing, the
canon might be intercut with RED; but with the teeth grinding
hope that it's brief enough and under conditions favorable enough
to not destroy the integrity of the RED footage.
But this does not agree with the spirit of these tests.

"Too sharp": there is no such thing as a too sharp lens/camera combo,
anymore than there is a "too in tune" musical instrument. Play it as you would.

 

[Graeme Nattress]

Oh I know, I was just correcting his post. Do you know what the measured resolution is of the 5D2/7D shooting video? Of course to be fair I remember hearing people complain that RED wasn't really 4K and that it was 3.2K. As far as measured resolution goes.

Personally I don't care because you can look at the images coming off of the 5D2/7D vs the RED and see there's a big difference.

On the 5D I found it hard to see more than around 1.3k and that was being generous. I've not had the pleasure of measuring the 7D in video mode. The GH1 was barely above SD in 720p and seemed to gain no extra resolution in going to 1080p.

Graeme

 

[marasco ivan]

do you have seen this using caprock filter? what do you think?
http://www.vimeo.com/7523222

 

[Sanjin Jukic]

do you have seen this using caprock filter? what do you think?
http://www.vimeo.com/7523222

Thanks shiny,

looks good.

I'll ordered it.

 

[Bryce O]

Yeah and I remember hearing that the DVX looked like 16mm. But is it even close...no. BTW the 5D2 and the 7D are 1920x1080 not 2K..

Of course the DVX doesn't look like 16mm. 16mm is 2/3" sensor. DVX is what? 1/3" inch? 7D looks like Red in the sense that they share the same depth and aesthetic qualities. Red has better image quality, but they "look" the same. That's why I thought the argument was dumb. Obviously, 4k is going to be sharper, but 35mm looks like 35mm. Cutting HVX or similar camera footage is defiantly not possible, as you are going from the deep 3D-like texture of 35mm to essentially a 8mm flat looking picture. That is what people will notice and that is why you can intercut Red, film, 7d, 5d, whatever...

I've shown this video to people with no clue about anything to do with cameras and, of course, they had no idea what was going on. They thought it was just a test for wipe transitions and coloring. Nobody had any idea it was about cameras until I told them. The best response was from a girl who said "that's a dumb test, they all look the same."

 

[Neil W. Smith]

The purpose of the RED/Canon Intercut reel ...

The purpose of the RED/Canon Intercut reel ...

I've shown this video to people with no clue about anything to do with cameras and, of course, they had no idea what was going on. They thought it was just a test for wipe transitions and coloring. Nobody had any idea it was about cameras until I told them. The best response was from a girl who said "that's a dumb test, they all look the same."

Thanks for the feedback, Bryce .... you've hit the nail on the head ... this wasn't actually a 'camera test' .... we already know how great RED RAW is and how much better the 4k .r3ds are to just about any other file format out there .... as a professional DI house we care fanatically about every pixel we grade .... this was primarily an exercise in seeing if a $1700 HD DSLR could hold up against a $17,000 4k digital cinema camera when projected in a professional DI environment .... I was totally skeptical before the shoot but when Rodney Charters, ASC, takes the Canon cameras seriously, you have to at least give them a whirl.

After the shoot and grading all three cameras in our highly calibrated DI environment we've become much more open minded about the possibility of intercutting well shot 5D and 7D with well shot RED ONE.

Given the choice, would we prefer to grade RED ONE footage over the Canon H.264? - you bet! .... but given the choice, we'd also prefer to grade Sony F35, Genesis and Arri D21 over the RED any day.

The point of the RED/Canon Intercut shoot was to see if "normal people" can easily tell the difference between RED and Canon footage that has been well shot, color corrected and screened in a decent digital projection environment ... 150 people at HD Expo and 100 or more so people at our Open House day (most of them professional photographers) were surprised how difficult it was to tell the difference .... even after we told them which camera shot what shot.

At the end of the day, what customers see up on the silver screen is all that matters ... do they stay in the 'willing suspension of disbelief' or do they suddenly think they're not watching a movie? ... we've had many ASC DPs in our theater who still swear blind that 35mm is the ONLY capture medium and hate the look of RED .... and yet, whenever we show them a well shot RED movie that's been graded by one of our talented colorists, they soon forget about the shortcomings of digital acquisition and concentrate on the aesthetics of visual story telling.

RED started the revolution in digital acquisition (I still have several of the very first 76 mg TIFFS off the first prototype RED ONE from way back in the summer of 2006) but there are several other large camera companies that have huge R 7 D departments that are serious about getting into this new wave of digital file based filmmaking.

Moore's Law drives everything in the digital revolution ... these new acquisition devices are nothing more than a computer with a sensor at one end and a piece of glass at the other .... we're going to repeat the RED/Canon Intercut Reel using Zeiss Master Primes .... be interesting to see what a decent piece of glass does for both cameras.

Neil

 

[Sanjin Jukic]

.... we're going to repeat the RED/Canon Intercut Reel using Zeiss Master Primes .... be interesting to see what a decent piece of glass does for both cameras.

Neil

Neil,

which Canon EOS DSLR PL mount did you use with Arri/Zeiss Master Primes???!!!

Just curious!!!??? :bigear:

 

[Miguel "Macgregor" De Olaso]

I see.

 

[Neil W. Smith]

Master Primes on RED and Canons ...

Master Primes on RED and Canons ...

Neil,

which Canon EOS DSLR PL mount did you use with Arri/Zeiss Master Primes???!!!

Just curious!!!??? :bigear:

Good question Sanjin ... the Zeiss guys approached us so I presume they'll have that figured out ... will let you know as soon as we've organized the next Canon/RED Intercut Reel.

For the last RED/Canon Intercut reel we decided that we would use RPPs for the RED ONE and Canon Glass for the EOS 5D Mk ii and EOS 7D ... both are mediocre glass and both give an equally bad image for movie making .... but we wanted to start at the Lowest Common Denominator (LCD) so we could see how much better it could get.

Even using LCD glass all three cameras produced pretty good looking images.

Using decent glass on the RED ONE and the Canon EOS 7D should give us a better comparison .... we'll also shoot some charts in a controlled environment so we can measure MTF values better.

Sanjin, what's your view on where digital acquisition and post is going?

Neil

P.S. my sister lives in Wien ... next time I come to vist her I'd love to come over and see your studio. N.

 

[Fredrik Callinggard]

Yeah and I remember hearing that the DVX looked like 16mm. But is it even close...no. BTW the 5D2 and the 7D are 1920x1080 not 2K.



That is true...and here's a picture to prove it. I found it interesting that you were using a 5D2 and an Arri film camera on the same rig. What was the reason you weren't using your RED cameras you have. I never got to ask.

Crew.... budget wouldn't allow for more people lol. Also I wanted the stills element that the 5D's provide, so we could shot better time lapse, long exposure stuff. Last one was that on the lakebed I used them as interior rigged cameras.... for space (Epic will most probably change that).

The size was a big factor for it. Secondly we used them for interesting long exposure stuff. One example was a one live action 35mm with 2. 5D's next to it. One with 1sec exposure and 1 with 1/4000 ... I will mix that together and have live action go into interesting freeze frame scenarios just to cut back out to live action (this was all the lake bed stuff, so you unfortunately missed that Andrew... a lot of nice models there :driving: female that is lol).

 

[Daniel Browning]

This discussion is stupid. Does 7D/5D footage "look" like Red One footage? Yes, if you disagree you are delusional.

Then I am delusional.

Is the the Red One sharper and better for grading? Yes, it's 4k RAW vs 2k compressed what did you expect. Aesthetically, they all look the same, 35mm format with decent dynamic range.

It's not just resolution. If the 5D2 was limited to 720p, but was correctly anti-aliased, then it would be 10 times better than it is now. Even 480p would be better than what we have now, if only it wasn't aliased.

To me, the 5D2 is very different aesthetically, precisely because of the aliasing. Aliasing makes any of the in-focus shots look like garbage. To me, aliasing gives it a harsh, fake, unnatural feeling. Anti-aliased images feel smooth, realistic, and natural. To each their own, of course.

The point of the RED/Canon Intercut shoot was to see if "normal people" can easily tell the difference between RED and Canon footage that has been well shot, color corrected and screened in a decent digital projection environment ... 150 people at HD Expo and 100 or more so people at our Open House day (most of them professional photographers) were surprised how difficult it was to tell the difference .... even after we told them which camera shot what shot.

For me, the only time that Canon footage looks similar to RED is when it is so blurry (out of focus, etc.) that there is not enough detail for aliasing artifacts to occur.

But then the discussion changes from "RED and Canon?" to "does resolution matter at all?". Here is a 100% crop from the 5D2 footage:



In the following image, I took the crop and resized that down to 360p, then up-rezzed it back to 1080p. 360p is 0.6 "K", which is 7 times less than "4K", 9 times less megapixels than 1080p, and even 50% *less* megapixels than Standard Definition DVD.



Looks pretty similar. That image had such little resolution that even Standard Definition is overkill.

What happens when we do the same thing with the RED ONE footage of the same scene? First the 1080p crop:



And here it is after resizing to 360p and back up to 1080p:



This time, the loss is striking. Clearly, the RED ONE shot had 2.1 megapixels of real detail in a 2.1 MP file. Whereas the 5D2 shot had about 0.23 MP of real detail in a 2.1 MP file.

But how do we reconcile this with the fact that normal people could not tell the difference? One response would be to decide that 360p is good enough, and SD cameras (480p) are overkill. If you take that route, you will decide that 480p, 720p, 1080p, 4K, etc. offer no advantage over 360p, since normal people can't tell the difference.

The other response is to continue in the delusional belief that resolution actually matters. That less-than-SD resolution is not adequate. That the detail offered by 720p and 1080p is truly important.

I don't think SD is overkill on resolution. If that makes me delusional, so be it.

 

[M Most]

But how do we reconcile this with the fact that normal people could not tell the difference? One response would be to decide that 360p is good enough, and SD cameras (480p) are overkill. If you take that route, you will decide that 480p, 720p, 1080p, 4K, etc. offer no advantage over 360p, since normal people can't tell the difference.

What you're saying is, of course, true. However, it is also true that the "normal" people Neil was referring to included a number of industry people and even a number of DP's. The fact is that when you're looking at stills, especially on a computer screen, you have a different reaction then you do when you're looking at moving footage on a larger screen. The many stills that make up that moving image tend to integrate, and a lot of things that are easy to point out on close inspection tend to be less obvious. And the additional fact is that regardless of the issues you and others here are bringing up, people are using these things for many projects, some of which you wouldn't expect them to consider them for, in part because of the very issues you're bringing up. And neither your statements nor anyone else's here are likely to change that fact.

Since all of the reaction here is remarkably similar to the early reaction of many experienced DP's and industry pros to the original introduction of Red, it's interesting to note that when Red was introduced, it was noted as costing about 1/10 the price of the existing high end cameras but likely providing a very high percentage of the image quality. The Canons can be said to cost about 1/10 the price of the Red and provide an acceptable percentage of the image quality (to many eyes, especially for HD video delivery), especially if one is looking at some of their specific advantages, such as low light performance. One conclusion might be that it's fun and strangely satisfying to be the disruptor. It's not fun to be the disrupted.

 

[Roberto Lequeux]

do you have seen this using caprock filter? what do you think?
http://www.vimeo.com/7523222

Wow... nice find!

It does make it far softer, but it it fine sometimes, certainly better than getting the LCD trip look that that last shot would have.

He also posted a detail loss test: http://www.vimeo.com/7561473

For the last RED/Canon Intercut reel we decided that we would use RPPs for the RED ONE and Canon Glass for the EOS 5D Mk ii and EOS 7D ... both are mediocre glass and both give an equally bad image for movie making ....

Do you really believe RPPs have mediocre glass, and give mediocre images? As far as I understand and can tell their image is stunning. Some would say the focus throw is too short, and the lenses are too big and heavy. But bad optical performance?

Thank you very much Neil for posting this comparison.

 

[Mark Pugh]

Using decent glass on the RED ONE and the Canon EOS 7D should give us a better comparison ....

I think you'd hardly even see the difference in optical quality between RPP and Masterprimes on an adapted Canon DSLR, given the huge loss of detail from line skipping evident in the 1080 frame grabs.
The difference in camera performance between the Canons and Red One far outweighs glass performance issues (at 1080 finish), aside of course from the fact that Masterprimes would be easier to focus if you were using the lens markings, partly overcoming a shortfall in the usability of the DSLRs (some accidentally soft images) that's complicated the results of this comparison test.

 

[Daniel Browning]

What you're saying is, of course, true. However, it is also true that the "normal" people Neil was referring to included a number of industry people and even a number of DP's.

I'd like to say I'm surprised, but now that I think about it, I've ran into many pro still photographers who feel that 2 MP is more than enough for 12x18, 20x30, and larger prints.

The fact is that when you're looking at stills, especially on a computer screen, you have a different reaction then you do when you're looking at moving footage on a larger screen.

Agreed.

The many stills that make up that moving image tend to integrate, and a lot of things that are easy to point out on close inspection tend to be less obvious.

That's true. Sometimes that works in reverse, though, and that's precisely the case with the 5D2's greatest flaw: aliasing. Aliasing in a still image is bad (very bad, IMHO), but at least they don't move. But in motion imagery, the aliases dance, flow, pop, strobe, sparkle etc. That makes them much, much worse.

And the additional fact is that regardless of the issues you and others here are bringing up, people are using these things for many projects, some of which you wouldn't expect them to consider them for, in part because of the very issues you're bringing up. And neither your statements nor anyone else's here are likely to change that fact.

Pfft. I can top that. Most of the stuff on my TV is already butchered to death by aliasing, compression, interlace twitter, and other junk. Some of my Blu-ray discs are over-sharpened to the point of ringing artifacts. Some of my film-going experience is tainted by issues there (aliasing on the FX comps, badly missed focus, etc.).

Nothing I say is going to change that fact, either. Compared to all that, the use of the 5D2 is a drop in the bucket.

when Red was introduced, it was noted as costing about 1/10 the price of the existing high end cameras but likely providing a very high percentage of the image quality. The Canons can be said to cost about 1/10 the price of the Red and provide an acceptable percentage of the image quality (to many eyes, especially for HD video delivery),

That makes sense. The only question is how you weight the importance of various aspects of image quality. Here's a few, for example:

• DOF control
• Contrast
• Low light
• Dynamic range
• Resolution
• Artifacts

If all you care about is DOF control and low light, then the 5D2 provides far higher image quality than the RED ONE. If you think SD video is overkill for resolution, then that aspect of the Canon DSLR is no disadvantage. If you actually love aliasing (and I think many people do), then the RED ONE is inferior to the 5D2, since it requires spending extra time in post to add the aliasing artifacts, whereas the 5D2 has them built in.

On the other hand, if you think the 5D2 aliasing is devil spawn sent to earth by Lucifer himself to torment us into clawing our eyes out, then you might hate the 5D2 with the passion of a thousand burning suns. You might prefer the lowly HV20 because it has higher resolution and less aliasing, even if it means the loss of DOF control and low light.

To me, aliasing sticks out like a sore thumb. It's like the difference between live action footage and 1990's computer animation. It smacks you upside the face with how fake and unnatural it is. Yet people are saying that you can intercut it with realistic and natural footage, like the RED ONE. The reason for the difference is that other people just perceive images differently than I do, even though we're both looking at the exact same image.

One conclusion might be that it's fun and strangely satisfying to be the disruptor. It's not fun to be the disrupted.

I'm really curious about what the result of the disruption will be. Will the success of DSLRs cause manufacturers to realize that all the years and billions of dollars of research they spent on systmes that avoid aliasing were a total waste? That it was actually counter-productive? That most of their customers desire the very aliasing that manufacturers fought to avoid? Will they finally start doing things like row skipping to give their customers large, cheap sensors with aliasing?

I'm curious to find out.

 

[M Most]

Will the success of DSLRs cause manufacturers to realize that all the years and billions of dollars of research they spent on systmes that avoid aliasing were a total waste? That it was actually counter-productive? That most of their customers desire the very aliasing that manufacturers fought to avoid? Will they finally start doing things like row skipping to give their customers large, cheap sensors with aliasing?

I'm curious to find out.

I think you, like a number of others here, are taking the presence of these devices way too seriously and way too literally. As I said before, you seem to be looking at this as an "either or" choice for an "A" camera. That's not an accurate assessment of the situation. While there certainly are those who will use them as a primary camera - probably a lot of the same people who flocked to the DVX100 when it showed up - for most professionals these things represent another tool, not a replacement for the main imaging device. I think you and others are spending a lot of energy trying to spread the very fear that those here have often accused others of spreading regarding limitations of Red - and Red does have limitations. So do DSLR still cameras appropriated for video use. But the fact is that Reds can be and are being used as the "A" camera, replacing film as well as other more expensive video cameras. And the Canons can be and are being used to do things that Reds don't do very well - like shooting available light night exteriors and 360 degree VFX plates. And if they occasionally get used in other ways, who cares? For a group that constantly screams about "democratization," isn't that exactly what these things are accomplishing, even if they don't happen to live up to standards that you personally deem acceptable?

 

[Daniel Browning]

I think you and others are spending a lot of energy trying to spread the very fear that those here have often accused others of spreading regarding limitations of Red - and Red does have limitations. So do DSLR still cameras appropriated for video use.

Fear is the mind killer, but it's also a great motivator. Fear leads to anger, anger leads to hate, hate leads to an improved camera. When others spread fear about the limited dynamic range of RED ONE compared to film, I think that's great! That's just the kind of fear mongering that will lead to better cameras in the future.

Of course, there are bad kinds of fear, too. Like fear of things that don't exist. For example, some were spreading fear that the RED ONE had poor resolution because of the Bayer CFA; that 1080p cameras were actually higher in resolution. That's like fear of the boogeyman, it's not real. I don't support the spread of that kind of fear.

So in the same way that people share that RED ONE has less dynamic range than film, I state that 5D2 has terrible aliasing. I don't think my posts really have the effect of spreading any fear. I've yet to meet a single person that I could convince to dislike aliasing. Either they love aliasing or they hate it, but they wont change just because I share my opinion about it.

But the fact is that Reds can be and are being used as the "A" camera, replacing film as well as other more expensive video cameras.

That's true, but it can require a change in how you do things. If you used to shoot for 11 stops of dynamic range with film, you may need to cut that down to 9 when using the RED ONE. Same with the 5D2: if you used to shoot for a clean 1080p with film, you'll need to cut that down to 480p if you want to avoid aliasing on the 5D2.

And the Canons can be and are being used to do things that Reds don't do very well - like shooting available light night exteriors and 360 degree VFX plates. And if they occasionally get used in other ways, who cares?

I care enough to post in a thread comparing the 5D2 and RED ONE. It's just like people mentioning the dynamic range limitations of the RED ONE. Some people think there is no limitation, because they only use 6 stops anyway. Others would like to use 11 stops, but can't because they are limited due shadows that aren't clean enough. To them, the dynamic range is important. In the same way, aliasing and resolution are important to me, so I talk about those limitations of the 5D2.

For a group that constantly screams about "democratization," isn't that exactly what these things are accomplishing, even if they don't happen to live up to standards that you personally deem acceptable?

They do have accomplishments in many areas for low-budget productions: DOF control, lens selection, color depth, and low light. That's why I use my 5D2 for quite a few projects now. But in many ways they are actually moving backwards in image quality: most notably in aliasing and resolution, where they are bested by three-year-old camcorders that sold for one sixth the price, like in this comparison of the 5D2 and HV10.

 

[Hrvoje Simic]

This is a great example what happens when you use stills camera for handheld motion imaging (ergonomics & weight distribution issues).

Check out the type of camera shake (full screen).

do you have seen this using caprock filter? what do you think?
http://www.vimeo.com/7523222

Sorry for drifting off topic.