RED vs. 7D/5D vs. HV/PX in NYC

[Ryan Patch]

Hey all -

So, I have searched the whole forum and web over, and I can't seem to find what I consider to be a fair test between the RED, a new Canon camera, and a Panasonic P2 camera, conducted by knowledgeable techs, shooting both test patterns and live-action subjects, and (this is the real sticker) processed well in post by people who know what they are doing with the format.

So, I've decided to conduct my own tests in NYC, aided by Vladimir K from vladlab, who is proficient in processing and coloring all three formats. The real emphasis will be on how much and well can we malipulate the image in post (color effects, shadow and hilight recovery, etc.) We all know that RED's strength is in it's RAW capture, as compared to the Canon's very compressed h.264. Of course the Canon can spit out pretty video, but how manipulatable is it? This is what most people don't understand, and I would love to have some solid data and visuals that demonstrate this.

I'm also including the HVX, because right now this is the standard for run-and-gun documentary shooting. The RED ONE is too large, and the 5/7D just doesn't have the audio options that are needed. I have had relatively good success with tweaking color, and want to compare it to the Canon cameras in a controlled environment.

I owned RED #2915 for one year, but sold when some international doc opportunities came up last year (see www.redlightgreenlightprods.com). As such, I need a RED camera for the testing in NYC, and I hope that I can find another reduser in the New York City area who is interested in helping us duke this out, and (hopefully) finding conclusively that the RED - and redcode - is truly the superior camera and format.

If you own a RED in the New York City area, and want to go out with us for a day, please let me know. You would be able to handle your RED, or if you don't have time, we can pick it up and take it out.

Thanks all! Please PM me, or contact me at d.ryan.patch at gmail.

Ryan

 

[Rob Ruffo]

Hey all -

So, I have searched the whole forum and web over, and I can't seem to find what I consider to be a fair test between the RED, a new Canon camera, and a Panasonic P2 camera, conducted by knowledgeable techs, shooting both test patterns and live-action subjects, and (this is the real sticker) processed well in post by people who know what they are doing with the format.

So, I've decided to conduct my own tests in NYC, aided by Vladimir K from vladlab, who is proficient in processing and coloring all three formats. The real emphasis will be on how much and well can we malipulate the image in post (color effects, shadow and hilight recovery, etc.) We all know that RED's strength is in it's RAW capture, as compared to the Canon's very compressed h.264. Of course the Canon can spit out pretty video, but how manipulatable is it? This is what most people don't understand, and I would love to have some solid data and visuals that demonstrate this.

I'm also including the HVX, because right now this is the standard for run-and-gun documentary shooting. The RED ONE is too large, and the 5/7D just doesn't have the audio options that are needed. I have had relatively good success with tweaking color, and want to compare it to the Canon cameras in a controlled environment.

I owned RED #2915 for one year, but sold when some international doc opportunities came up last year (see www.redlightgreenlightprods.com). As such, I need a RED camera for the testing in NYC, and I hope that I can find another reduser in the New York City area who is interested in helping us duke this out, and (hopefully) finding conclusively that the RED - and redcode - is truly the superior camera and format.

If you own a RED in the New York City area, and want to go out with us for a day, please let me know. You would be able to handle your RED, or if you don't have time, we can pick it up and take it out.

Thanks all! Please PM me, or contact me at d.ryan.patch at gmail.

Ryan

The differences are well known. Many people have shot with both. Including me.

The 5D comes up lacking in 3 main areas:

- Low actual resolution: far less than the almost perfect 1080p an EX1 will give you, and even further less than the 3+K of actual resolved pixels a Red will give you after debayer. Resolution has been measured, over and over, as just in the middle between 480p and 720p. This is not up for debate. My own eyes tell me this is true.

- Moire moire moire!!!! Moire everywhere! Moire on fabrics, moire on teh lines in people's foreheads, moire on bricks, tiles (basically forget about shooting any buildings in focus.) Even their own demo film form the MK IV (which uses teh same process for reducing sensor size), Nocturne, has Moire all over many building facades. This si not simple line-aliasing - this is psychedelic rainbows that absically trash your shot in many cases.

- Serious skew. Way, way more than Red or EX1.

5D wins in three areas:

- Better low light response.

- Easy to get look you want in-camera, so less need for grading (but not much possibility of grading either - that codec is BRITTLE).

- Less dynamic range than EX1/EX3 but slightly more than Red (without Mysterium X)

 

[Álex Montoya]

That thing with the dynamic range you'll have to prove to me.

 

[Gary Stone]

watch district 9. not a canon 5D.

 

[JanneJansson]

I recently used 7D, HVX200 and a R1 on a shoot.
For SD, HD720 work they match pretty good. Just being careful with the light temp and settings on all cam and then no problems The most difficult camera to control is the 7D. R1 and HVX are workhorses that just keep goin. The 7D is far better in low noise, but also most tricky to work with.

Not doing "video" work with the 5D. Lack behind the 7D in alias/quality at the moment IMO. Maybe the new firmware will fix that.

 

[Dan Hudgins]

Grading range

Grading range

Grading range of H.264 is vary small maybe +/- 1 one half stop before the compression artifacts start to show from increased contrast.

Thats what grading is, you increase part of the curve(s) contrast and reduce the contrast in other parts. H.264 has just enough detail if you don't make any changes to look usable from a good distance.

If you get closer using full size images, and boost part of the curve it goes down hill.

RED CODE has a better range as far as I have seen in my tests, but still has issues in the deep shadow areas over maybe true RAW cameras. With True RAW data you get noise that is more like film grain rather than wavelets or blocks.

Acam dII is set to come out soon with its Kodak sensor being about EI/ISO 320, and can push to maybe 1280.

The 7D has the advantage of a large sensor and very heavy digital processing including some kind of DRX processing, you can see the luma hold in over exposed areas and the chroma fades out to nothing.

The digital processing of the sensor data makes for maybe 80% of the camera's look, most un-processed sensor data has quite a bit of noise even with near-over-exposure.

What sets the cameras apart is how the maker deals with the native noise in the sensor.

If you sharpen in post or boost the contrast or saturation then you un-do some of the tricks used to hide the sensor noise.

So its hard to compare cameras that are not ment to match in their "one light" output level, some may grade better than others, RED CODE will probably beat any Video or DSLR movie mode because was desigined to be graded, the others are ment to be viewed as is.

Beacuse of that if you shoot with getting the exposure right in the camera, then grade RED ONE to match the look of the 7D then all the cameras will stand up well, but if you light for the most extream save of the RED ONE in post grading and noise filtering (neat video) then the others will have problems since they will not let you do anything other than adjust their in-camera settings, and the lighting may be out of the control range of the in-camera settings.

Its hard to come up with a fair test, say light for "best" results in all cameras, rather than pick a random lighting situation and see "what happens"?

 

[JanneJansson]

..but, despite that these cams are so different, they match pretty close if you are careful with white balance and cam settings.

I just removed a bit chroma from the 7D, and it matched perfectly to HVX.

 

[MichaelP]

Grading range of H.264 is vary small maybe +/- 1 one half stop before the compression artifacts start to show from increased contrast.

Dan - just want to be sure as to what you are saying here: Are you saying a total of 1.5 stops over and under?

Thanks - Michael

 

[Rob Ruffo]

That thing with the dynamic range you'll have to prove to me.

The dynamic range of 5D video is only slightly better than Red's, but the roll-off is predictable and tends toward less surprises. Color definition also seems better to me, allowing for a natural, still pelasant high-sat look, but no scientific tests on this.

Don't get me wrong, I avoid shooting video on my 5D for reasons stated, but it does have a few strong points, and it's important to give credit where credit is due.

 

[Rob Ruffo]

Dan - just want to be sure as to what you are saying here: Are you saying a total of 1.5 stops over and under?

Thanks - Michael

No, he means 1/2 stop in either direction. I'd say it depends what, but smooth gradients are the first to lose acceptability, and of course a well lit human face is a smooth gradient so....

 

[Rob Ruffo]

I recently used 7D, HVX200 and a R1 on a shoot.
For SD, HD720 work they match pretty good. Just being careful with the light temp and settings on all cam and then no problems The most difficult camera to control is the 7D. R1 and HVX are workhorses that just keep goin. The 7D is far better in low noise, but also most tricky to work with.

Not doing "video" work with the 5D. Lack behind the 7D in alias/quality at the moment IMO. Maybe the new firmware will fix that.

The aliasing on 7D and 5D are the same.

 

[Miltos Pilalitos]

The dynamic range of 5D video is only slightly better than Red's

That's the first time i read such a bold statement. Are there any links that demonstrate this?

 

[Ryan Patch]

Are there any links that demonstrate this?

Getting back to the point of my original post, I would love to do standardized, controlled tests to prove this. Any NYC Redusers going to help me out?

 

[Kholi Hicks]

The whole idea that 7D and 5D are better in lowlight than RED ONE has actually become a falsehood, to me. Having extensive amounts of time with the 7D and 5D and, of course, shooting RED constantly, it's become very apparent to me that this claim comes from a "look once" basis. It's all dependent on which glass you're mating the RED with, of course.

First, the 7D and 5D are pretty grotesque above 640 ISO, especially the 5D. The fixed patterned noise (cross-hatching below the surface of the image) and "oatmeal" texture-- to borrow from Barry Green before anyone else does, haha-- render the image disgusting, only internet display worthy. Once you put an image shot at 640, even, on a large screen it begins to look very "ugh."

This is for lowlight, contrasty imagery. If shooting for a large display, Hi-Definition distribution, going above 320 ISO is very chancy, I wouldn't dare recommend it unless it was a MUST, and then it would only be 640 ISO.

With that in mind, compare the image of a current RED ONE setup (not projecting on an M-X upgrade) mated with Master Primes Wide Open (T1.3) @ 500ISO to a 7D strapped with a Zeiss F1.4 lens @ 320 ISO.

The RED's image not only looks, in many ways, "better" but it's also going to see more.

There is no versus: Anything sub RED ONE is exactly what it is, SUB RED ONE. Even without an M-X upgrade there is still no comparison, an M-X upgrade is simply going to eat many more lunches.

Then again, it also costs a lot more... which doesn't really matter, honestly. But, that's another discussion?

 

[Álex Montoya]

Originally Posted by Rob Ruffo
The dynamic range of 5D video is only slightly better than Red's

-----------------------

I read an analysis on the spanish magazine 'Cameraman' that stated that while the 5D showed 8+ usable stops in its still mode, the video image was so processed and contrasted that it only offered about 5 stops of usable latitude.

 

[Rob Ruffo]

The whole idea that 7D and 5D are better in lowlight than RED ONE has actually become a falsehood, to me. Having extensive amounts of time with the 7D and 5D and, of course, shooting RED constantly, it's become very apparent to me that this claim comes from a "look once" basis. It's all dependent on which glass you're mating the RED with, of course.

First, the 7D and 5D are pretty grotesque above 640 ISO, especially the 5D. The fixed patterned noise (cross-hatching below the surface of the image) and "oatmeal" texture-- to borrow from Barry Green before anyone else does, haha-- render the image disgusting, only internet display worthy. Once you put an image shot at 640, even, on a large screen it begins to look very "ugh."

This is for lowlight, contrasty imagery. If shooting for a large display, Hi-Definition distribution, going above 320 ISO is very chancy, I wouldn't dare recommend it unless it was a MUST, and then it would only be 640 ISO.

With that in mind, compare the image of a current RED ONE setup (not projecting on an M-X upgrade) mated with Master Primes Wide Open (T1.3) @ 500ISO to a 7D strapped with a Zeiss F1.4 lens @ 320 ISO.

The RED's image not only looks, in many ways, "better" but it's also going to see more.

There is no versus: Anything sub RED ONE is exactly what it is, SUB RED ONE. Even without an M-X upgrade there is still no comparison, an M-X upgrade is simply going to eat many more lunches.

Then again, it also costs a lot more... which doesn't really matter, honestly. But, that's another discussion?

My commenst were re: no X sensor upgrade. I have no experience with X-sensor, and it looks fantastic. I stand behind what I said though, about a standard R1 sensor and the 5D - I don't own a 7D but I hear it is less good in low light than a 5D, so you're probably right about that cam.

 

[Rob Ruffo]

Originally Posted by Rob Ruffo
The dynamic range of 5D video is only slightly better than Red's

-----------------------

I read an analysis on the spanish magazine 'Cameraman' that stated that while the 5D showed 8+ usable stops in its still mode, the video image was so processed and contrasted that it only offered about 5 stops of usable latitude.

This is simply not true, in practice, with a well-tuned pic profile. The 5D gives you about 9 - 9.5 stops. AN EX1 gives you about 11, with their standard firmware and a well-tuned pic profile, with care in post not to clip 235-255 values.

I am NOT defending the 5D here. It is a very stupid choice in many circumstances. RED is worth the extra money, many times over. I am only saying that, sans a sensor X-upgrade, it still holds a few advantages in some cases, FOR NOW.

 

[JanneJansson]

The aliasing on 7D and 5D are the same.

Nop "IMO"

 

[A. Clint Litton]

Getting back to the point of my original post, I would love to do standardized, controlled tests to prove this. Any NYC Redusers going to help me out?

I'd be willing to help you out, Ryan, depending on scheduling availability of course. I was trying to set up a few 7D/Red side-by-side comparison tests (resolution, latitude, as well as some subjective tests), ideally using Zeiss ZE primes and a Canon mount for the Red, but was unable to source the mount here in New York. A test is still in the works somewhere, but will likely need a Nikon mount (sold mine, unfortunately), as it would seem very useful to use the same glass between the cameras.

Either way, shoot me a PM or email (clintlitton at gmail dot com) and we can figure something out.

Cheers,
Clint

 

[noirist]

Ryan,

How do you plan to test the three cameras?