Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Dragon vs Canon 1Dx Color Matching

Zeb B

Zeb B

Well-known member
Joined
Sep 30, 2011
Messages
4,933
Reaction score
1
Points
0
Location
In The Moment
So I'm out shooting some wildlife with me Mate . . . he just happened to click off a 1Dx frame at the same moment and similar angle as what I was filming. Back at base camp we open both up and his CR2 looks great out of camera. No surprise there since that's the Canon thing. The noise difference and the work it takes to get my R3D to look anything like the attached JPG (accurate brown on the Elk) is surprising. The shadow on the neck is a good place to key in on. The color cast in the R3D is persistent and efforts to kill it seems to compromise the other colors. Ideas? What did I fuq up?

All images ©ZEB

CR2 - shot with 70-200mm
https://dl.dropboxusercontent.com/u/117794729/Dragon Canon Elk/_EB_0312.CR2

R3D - shot with Leica Summicron-C 100mm with Motion Mount wide open and lens around T8
https://dl.dropboxusercontent.com/u...on Elk/A042_C015_0527DZ.0000585F.0000000F.R3D

CR2 after basic LightRoom treatment:
https://dl.dropboxusercontent.com/u/117794729/Dragon Canon Elk/©ZEB_Elk_LR-214.jpg
 
Saturation doesn't look right on Dragon. Your friend also shot with a higher shutter speed and lower ISO (Cleaner image and sharper image from lack of motion blur). He apparently shot with a t6.3 and you shot with a t8 so his shallower depth of field will also give the appearance of a sharper image. The 1DX shot looks excellent to me. Something about Dragon makes me wonder if the Motion Mount isn't playing right with the newest RCX. The colors just don't look right at all. No adjustment really "fixes" the color cast so to me I'd start diagnosing the issue. Try a shot in a similar area with and without the motion mount and see if that makes a difference. I boosted saturation all the way up on Dragon and it still wasn't saturating much. Also try shooting at a lower compression ratio as high detail scenes like this cause compression artifacts to heavily degrade your image. When theres lots of trees, foliage, and detail, aim for an 8:1 or lower compression ratio on Dragon. For wildlife stuff definitely up the shutter speed.

To me, something is up with this image and I hope it's a software thing because it doesn't look right to me at all for Dragon (but I've never used the motion mount). This 1dX shot is stunning and Dragon should look similar. My guess is it's related to the motion mount...:
1dxshot.jpg
 
See if this looks better on the Red shot:



This is a trivial, trivial match in Resolve. I've had far worse shots to match, particularly when one of the cameras is misadjusted. I would never, ever try to judge camera quality just on the basis of an uncorrected file like this.

Update: I see the o.p. took down the original posted Red image, but this is a better match now to the 1D.
 
The Dragon shot looks slightly over exposed to me, although I can only see Marc's version and the tail seems to be closer to clipping and the sheen of the fur and antlers seems brighter.

The main problem with the comparison though is that the Canon looks nicer due to having a shallower depth of field and larger field of view. Cropping them to match field of view helps to be judgmental.

Do yourself a favour, when testing a stills camera against the epic use the exact same lens at the exact same shutter speed. If you were shooting at 24p with a 180% shutter then the Epic is exposed for 1/48th of a second. I doubt the Canon would've been exposed for that length of time because it doesn't want the motion blur.

My point being that you can't compare anything without matching them to as exact spec as possible.

Paul
 
Hey Guys,
Thanks for the input. I usually shoot this type of work at 48fps/100th. The Compression minimum at 6K FF / 48fps is 10:1 with the non-mini SSD's. If 10:1 is crap - then we got a serious problem since that is as low as she goes and I'm not looking to drop another $20K on memory related gear just to get a clean image at 48fps.

The shadow noise when I'm ETTR to the point of clipping is something I'm not understanding. I mean it's just making up colors in the Elk's neck.

I'll do some testing to eliminate some variables. Not what I was expecting to have to do since the work seems pretty straight forward and the gear is 5X the cost of Canon. This was never intended to be a test comparison but looking at that day from both cameras it's definitely advantage Canon for RAW color and fidelity out of the box. I'm hoping that the Motion Mount is creating the issues. I'm having a hard time bonding with that piece of gear.
 
Marc Wielage said:
See if this looks better on the Red shot:



This is a trivial, trivial match in Resolve. I've had far worse shots to match, particularly when one of the cameras is misadjusted. I would never, ever try to judge camera quality just on the basis of an uncorrected file like this.

Update: I see the o.p. took down the original posted Red image, but this is a better match now to the 1D.
Click to expand...

I didn't take down anything. Please explain your "misadjusted" statement so I can learn from the experience. It seems your color correction simply crushes the detail out of the neck area and therefore "fixes" the major noise issue. Look closely at the beard / hair fringing and you'll see there is a huge difference in color fidelity and detail between the Canon shot and the RED. It's not DOF, Shutter Speed or F-Stop difference. Maybe the moving grass detail is breaking the wavelet math and the required 10:1 compression is not a working ratio.

I'm not trying to start a R vs C thing here but I just stood 30' from a 1000lb Alpha Male to get the shot of a lifetime . . . not to do a noise test.

If this chroma noise and soft detail is normal for Dragon then fine - I just need to know what I can count on when the shot is critical. If anyone else have a similar shot to this and is willing to post the R3D with shot notes then I can verify what I am seeing is or is not normal.
 
Zeb B said:
Hey Guys,
Thanks for the input. I usually shoot this type of work at 48fps/100th. The Compression minimum at 6K FF / 48fps is 10:1 with the non-mini SSD's. If 10:1 is crap - then we got a serious problem since that is as low as she goes and I'm not looking to drop another $20K on memory related gear just to get a clean image at 48fps.

The shadow noise when I'm ETTR to the point of clipping is something I'm not understanding. I mean it's just making up colors in the Elk's neck.

I'll do some testing to eliminate some variables. Not what I was expecting to have to do since the work seems pretty straight forward and the gear is 5X the cost of Canon. This was never intended to be a test comparison but looking at that day from both cameras it's definitely advantage Canon for RAW color and fidelity out of the box. I'm hoping that the Motion Mount is creating the issues. I'm having a hard time bonding with that piece of gear.
Click to expand...

The REDCODE compression isn't really a factor in your antler shot here. It's more of a combination of a different lens and a different recording medium. Canon has a fairly aggressive curve and they bake their color in their own unique way. Color-wise the blue channel is the real tricky one here and the lens itself is causing additional colorful effects that can be tackled one by one if this was an actual match job. Without creating masks I can get it to here:

phfx_zebAntlersCanonDragon.jpg


Though the antlers are getting a bit more of the ambient skydome blue fill light in the Canon shot. This could be due to a million reasons. Using the exact same lens (not a different copy of the lens, the same exact one) and the same shutter speed and aperture would minimize some of that chaos. I'm familiar with both the lenses on display here and they do render color and detail differently. Angle of incidence is slightly at play here too, but that's pulling hairs and mainly is reflected on the sheen in this particular shot.

Overall remember that Canon in particular is doing lots and lots of crazy stuff via their Digic processors. This is reflected in their motion and still images in the form of noise reduction and sharpening mainly (as well as processing speed). Red does not employ any in camera sharpening or noise reduction and there is indeed a reason for this. Also the Motion Mount is adding an additional variable here, something that really can't be taken into account when it comes to comparing against any other system not using it.

I'll take a close look at your boat clip in a bit.
 
Last edited:
Thanks for the effort and explanation Phil!

Forget Canon for a second - just wondering if what you see in Noise / Detail both here and on the Boat is normal for Dragon.
 
Phil Holland said:
The REDCODE compression isn't really a factor in your antler shot here. It's more of a combination of a different lens and a different recording medium. Canon has a fairly aggressive curve and they bake their color in their own unique way. Color-wise the blue channel is the real tricky one here and the lens itself is causing additional colorful effects that can be tackled one by one if this was an actual match job.
Click to expand...

Phil Holland said:
Though the antlers are getting a bit more of the ambient skydome blue fill light in the Canon shot. This could be due to a million reasons. Using the exact same lens (not a different copy of the lens, the same exact one) and the same shutter speed and aperture would minimize some of that chaos. I'm familiar with both the lenses on display here and they do render color and detail differently. Angle of incidence is slightly at play here too, but that's pulling hairs and mainly is reflected on the sheen in this particular shot.

Overall remember that Canon in particular is doing lots and lots of crazy stuff via their Digic processors. This is reflected in their motion and still images in the form of noise reduction and sharpening mainly (as well as processing speed). Red does not employ any in camera sharpening or noise reduction and there is indeed a reason for this. Also the Motion Mount is adding an additional variable here, something that really can't be taken into account when it comes to comparing against any other system not using it.
Click to expand...

I do agree that all of these things COULD be a factor.

However you must also agree, that there is another MORE LIKELY factor: RED's sensor and color processing combo is lacking compared to a good Canon DSLR photo.

Why don't you just go and say it. It's not like anyone will think you're anti-RED or something. You've done a massive amount of work for the community.

In order to get those two to match closely, you had to make the Canon shot look a ton worse to match the RED. Whereas what we want is for the RED stuff to be able to made to look like the Canon stuff.

Bruce Allen
www.boacinema.com
 
Bruce Allen said:
I do agree that all of these things COULD be a factor.

However you must also agree, that there is another MORE LIKELY factor: RED's sensor and color processing combo is lacking compared to a good Canon DSLR photo.

Why don't you just go and say it. It's not like anyone will think you're anti-RED or something. You've done a massive amount of work for the community.

In order to get those two to match closely, you had to make the Canon shot look a ton worse to match the RED. Whereas what we want is for the RED stuff to be able to made to look like the Canon stuff.

Bruce Allen
www.boacinema.com
Click to expand...

Indeed.

Even my 12mp Nikon D3s blows Dragon out of the water. same lens. Red's strong olpf isnt really suited for stills mode.
Monochrome on the other hand looks great.
 
Bruce Allen said:
I do agree that all of these things COULD be a factor.

However you must also agree, that there is another MORE LIKELY factor: RED's sensor and color processing combo is lacking compared to a good Canon DSLR photo.

Why don't you just go and say it. It's not like anyone will think you're anti-RED or something. You've done a massive amount of work for the community.

In order to get those two to match closely, you had to make the Canon shot look a ton worse to match the RED. Whereas what we want is for the RED stuff to be able to made to look like the Canon stuff.
Click to expand...

I matched the Red to the card version of the CR2. Didn't do any "worsening" as you described.

There's also nothing to "go ahead and say". I know exactly what the Dragon sensor is and I feel the color science is pretty darn good right now. Canon makes good sensors too and have the advantage of "rolling their own". We are just looking at two different things. And to be clear I used to be a Canon CPS member and still use a litany of Canon cameras for still work.

The fingerprint of the lenses is clearly visible in the images and that's that. The two systems in general create images in a slightly different way. Red is a motion picture camera first, while the Canon is focused on creating the best still image. Various optics, low pass filter, and internal hardware will prevent these systems from truly being the same.

As for the blue on the antlers I have a sneaky suspicion that the linear polarized Motion Mount is cutting it out passed on that angle perhaps. I would need to do controlled tests to be sure.
 
Zeb,
I will say those Summicrons are super disappointing. Not as sharp as I expected, lots of CA in examples I've seen (Some are from your tests). Not impressed at all. How do you feel about them?

I agree, color is off on Dragon here and can't pinpoint it. The noise does not look normal but I'd probably be shooting 320ISO in those situations.

Even pushing Dragon around hard to match, theres still a slight magenta cast in the fur and the green grass is off. This is what Dragon should look like...

1dxshot.jpg


Compared to Marc's correction:
Deerfixed_1601_zpsab8f2d36.png


Even Phils correction is off on Dragon (Look at the grass and the fur.):
phfx_zebAntlersCanonDragon.jpg
 
Zeb B said:
So I just need verification that the noise you see in the shadows on the boat at 200% and detail resolution is normal for Dragon:

https://dl.dropboxusercontent.com/u/117794729/Dragon Canon Elk/A040_C012_05208G_001.R3D

No Motion Mount, Leica Summicron-C, New OLPF installed correctly
Click to expand...

That doesn't look right either. It looks soft (lacking any real sharpness, and the noise looks pretty bad on the lower side of the boat. A complete lack of detail there as well. You might want to file a support ticket and have it looked at. It just doesn't look right to me. I even used ADD with heavier sharpening on this shot...
BoatZeb.jpg
 
Zeb B said:
I'm not trying to start a R vs C thing here but I just stood 30' from a 1000lb Alpha Male to get the shot of a lifetime . . . not to do a noise test.
Click to expand...
ROFL.

If the match is important enough to worry about, doesn't that basically disqualify the 1Dc as a B-camera in this situation? Why hold your Epic to a lower standard?
 
Zeb B said:
I didn't take down anything. Please explain your "misadjusted" statement so I can learn from the experience. It seems your color correction simply crushes the detail out of the neck area and therefore "fixes" the major noise issue.
Click to expand...
OK, you're welcome.
 
Had a play in Redcine-X, then fiddled a bit more with the contrast, saturation and sharpening in the photo-viewer I was using.


Canon (colours unadjusted) -



Dragon -





Canon (colours unadjusted) -



Dragon -





Kind of similar I guess. Not sure how you'd get the Dragon shot to look like the LightRoom adjusted Canon shot though.
 
Matt Ryan said:
That doesn't look right either. It looks soft (lacking any real sharpness, and the noise looks pretty bad on the lower side of the boat. A complete lack of detail there as well. You might want to file a support ticket and have it looked at. It just doesn't look right to me. I even used ADD with heavier sharpening on this shot...
Click to expand...

FUQ!!

I appreciate your candid and experienced insights. Problem is that my camera was at RED recently to flip the OLPF. I asked the tech there to double check the resolution / noise since I wasn't getting blown away by either. Seems that my old MX was sharper and cleaner. Was told that what I am seeing is normal Dragon behavior.

I'm sure if what we are looking at here is not typical Dragon performance then someone at RED will chime in here to state otherwise and what the fix is. I have no idea what I could have done different and only want to know how to fix it quick. I'm on a project that is time bounded.

Seems that the new-new OLPF, 10:1 trying to compress highly textured / dynamic scenery might be the smoking gun here. That and a wonky Motion Mount . . . which is being put in time-out until it learns how to play with others.
 
Phil Holland said:
I matched the Red to the card version of the CR2. Didn't do any "worsening" as you described.

There's also nothing to "go ahead and say". I know exactly what the Dragon sensor is and I feel the color science is pretty darn good right now. Canon makes good sensors too and have the advantage of "rolling their own". We are just looking at two different things. And to be clear I used to be a Canon CPS member and still use a litany of Canon cameras for still work.

The fingerprint of the lenses is clearly visible in the images and that's that. The two systems in general create images in a slightly different way. Red is a motion picture camera first, while the Canon is focused on creating the best still image. Various optics, low pass filter, and internal hardware will prevent these systems from truly being the same.

As for the blue on the antlers I have a sneaky suspicion that the linear polarized Motion Mount is cutting it out passed on that angle perhaps. I would need to do controlled tests to be sure.
Click to expand...

Thank you Phil. Could you state if the R3D's of both Elk and Boat is what can be expected from a Dragon in resolution, noise and color rendering given the scene / setting parameters? Just need to know if I have a broken camera or misplaced expectations

Thank you Les - beautiful grades as always
 
You must log in or register to reply here.
Back
Top