Why the 7D doesn't look so cinematic as the 5D?

[Chris Barcellos]

Originally Posted by Ian G. Thompson

Wow Chris....now that is an amazing eyeopener. I'm not an expert...but from what I am reading on those charts you posted it clearly shows around 800 lines of resolution. That's strange because Barry's test shows lower. But I think you might have found something here...which might explain why there are some videos I've watched that appears to be a heck of a lot sharper (or more detailed) than others....even using similar lens. Using superflat a "0" sharpness really seems like it helps.

I posted this on another thread, this is the the picture style file that was developed by another user. Sorry I don't remember who to give credit too. You need to change the extension to .pf2 and load to one of the user style through the EOS Utiltity.

http://www.dvinfo.net/forum/attachme...uperflat01.doc

[Jon Fairhurst]

Originally Posted by Cris Daniels

First because you are barely shooting HD with the camera, it would be a complete lie to tell a client they are getting true 1080p footage ...

Why would this even come up, unless you're shooting for an engineer?

A friend of mine shoots professionally with the 5D2 and various rentals. When pitching the 5D2, he shoots some footage and plays it back into an HDTV over HDMI directly from the camera. That usually makes the sale.

[Ian G. Thompson]

Originally Posted by Chris Barcellos

I posted this on another thread, this is the the picture style file that was developed by another user. Sorry I don't remember who to give credit too. You need to change the extension to .pf2 and load to one of the user style through the EOS Utiltity.

http://www.dvinfo.net/forum/attachme...uperflat01.doc

Thanks. I'll try this once I pickup my cam on Monday. It's the 7D but I understand that these profiles can be used on either machines.

[Chris Barcellos]

Keep us posted about what kind of results you see.

[Cris Daniels]

I dont know how to respond to comments like "how I don't touch these cameras" when I own and shoot them both as I've said ad nauseum.

Certainly shooting test charts isn't the be all end all, but it is hardly pixel peeping. I use the DCS Labs charts
ChromaDuMonde
Resolution B/W

The fiddlehead and backfocus chart is more for use with the video cameras since many allow you to adjust backfocus. The only reason I use this chart with the Canon SLR's would be to compare lenses or something like corner sharpness. Sometimes it helps to take the subjective element out of the loop. As long as you dont live and die by test charts, I dont see the problem with using them to obtain real information. To me its is just as valid as shooting a sample grid for lens distortion when compositing 3D to live action plates via match moving. You need to know your lens characteristics. If this is all "pixel peeping", well so be it. Having all of these things sorted out allows me to concentrate on shooting and knowing exactly what will work best for me when I encounter a given situation.

I shot with the 7D all morning as a matter of fact and it was because of the focal length. I needed over 1000mm and this camera is the only one I have that can do that.

I won't even debate the image quality versus a real video camera, and the comment that I work for the competition is laughable! I am an end user, and I love Canon but I am not a cool-aid drinker. If the camera exhibits certain issues, I call it like it is.

I will continue to use these cameras but never as a substitute for a real video camera. Heck a Panasonic HMC-150 for $3600 is a very nice little HD camcorder if you want something in that price range. Its not like the only option is the 7d/5d or $20k plus.

Also, I'm just not interested in arguing about any of this, I am too busy and it is a waste of energy for everyone. I will post some videos at a new vimeo account and link back to them. Thats all. Ultimately to each their own, and if a DP or director wants to shoot with a 7D or 5D as an A cam, more power to them.

[Chris Barcellos]

Cris Daniels: Have you tried the flatter settings with lower sharpness settings as I suggested, to improve resolution ?

[Cris Daniels]

Here is one video I had to dug up. I have another one but I will upload it in the morning. You really need to download the file, viewing this stuff on the net is worthless.

5D Moire Example on Vimeo

I have had the sharpness at +1 which is almost off, and the contrast totally flat. I don't use anything other than the built in camera curves. Although the sharpness settings would not influence the moire I am showing here, if anything the shot is soft but the lighting was bad as we had already unplugged the ARRI's to let them cool down so I could pack them. I was only working off available light in a bar, This was either ISO 1600 or 3200 @ f2.8, 1/60th shutter.

The shot is a turd but it does show the moire jumping around. Dont judge the overall quality of the video, the camera can certainly do better, but this is pointing out a specific issue which is unrelated to the quality of the shot.

My next shot will show the wacko aliasing. Usually I delete these kinds of files so it took me a while find some on backup drive.

I shot all morning with the 7D and the footage looks good, sometimes water gives me trouble but not today. I seem to have the most aliasing trouble with water in the shot flowing diagonal or top to bottom versus side to side. And no I can't always just move the camera.

[Barry Green]

Originally Posted by Ian G. Thompson

Wow Chris....now that is an amazing eyeopener. I'm not an expert...but from what I am reading on those charts you posted it clearly shows around 800 lines of resolution. That's strange because Barry's test shows lower.

When you're reading a chart, you don't go looking for the smallest area where you can make lines out, you look for the point at which the camera quits representing the chart accurately.

For example -- look at the diagonal lines that are about 2/3 of the way over on the right, the vertical column of 1/3/5/7/9/10/8/6/4/2. Those should be a patch of diagonal lines. The ones that are accurately rendered are 1, 2, 3, and 4. 5 is okay-ish, you can clearly see that there's aliasing happening but it's at least still a rendering where the black and white lines are cleanly separated. In the "6" box you can see the aliasing pattern clearly; where the lines are supposed to be going from upper-left to lower-right, there's a cross-pattern being introduced that makes the whole box look like diamonds. That is not an accurate representation. The camera did not successfully represent what the chart looks like in the "6" box. "7" is completely gone, as are 8, 9, and 10. So you know the real resolution being seen here is somewhere below 600. If the resolution was over 600 lines, that box of diagonal lines should have been cleanly represented, and it wasn't.

For more, look at the row of angled mostly-vertical lines along the bottom. The boxes with a slight lean, from upper-right down to lower-left. Here the last box you can see that looks "right" is the "5" box, the "6" looks funky, and "7" is clearly fuzzed (can something be "clearly" fuzzed?) So, again, pointing to a true res of around 600.

In pure horizontal and pure vertical, you should see a uniform pattern of distinct lines. It looks like in the horizontal lines (the big vertical column about 1/3 of the way from the left) 400 is solid, 500 is arguably pretty good, 600 is pushing it. 600 definitely looks like there's some thicker and thinner lines in that patch. In the pure vertical, which is the row of boxes at the bottom of the chart, you should again see a clearly delineated series of equal-width lines, alternating between black and white, all of the same thickness. Based on that, it starts to look a little "off" at about 550; and by 650 it's clearly not accurately representing the chart anymore.

Okay, so moving on to the resolution trumpets -- these are the wedges in the center and corners of the charts. What you want to see there is a smooth flow of lines with no oddness, they gradually converge. And you're looking for the point where the camera has not accurately represented what the chart looks like. It's not a point of looking to see where you can distinguish any difference from black to white, you're looking for accurate representation. And based on that standard, I'd say the horizontal line trumpets (in the center of the chart) are showing cleanly up to about 550, where you start to see some aliasing kicking in and making stair-steps on the lines. It's hard to see where the point is where it first fails to accurately represent the chart, but my guess is, once again, about 650. If you look at the horz line trumpet in the upper right corner it's a lot easier to see that it can't accurately represent the chart at 650.

The vertical trumpets are of interest are the ones near the top of the chart, just right of the center. The camera does fine with the 100 to 600 trumpet to the left of center, so the one to the right of center is where we'll find the maximum point. And once again, I'd say it's at about 650.

The tough thing about using an ISO12233 like this is -- where did it come from? Is this the downloaded/printed one? Because if so, how high was the resolution on the printer? If the printer doesn't have enough res to accurately represent those finer lines, then it's not too useful as a true indicator of resolution. When I was using one of these, my tactic was to print it at 4x size, that way it helps remove the printer inaccuracies from the equation.

Nothing beats having a properly-designed res chart such as the DSC Labs Chroma Du Monde with resolution trumpets, or the MegaTrumpets 4K (which is way overkill for DSLRs, since the chart starts at 500 lines and the DSLRs can barely register on the chart before they've aliased out.

Finally, look at the color version of Chris's chart again -- the black and white one lets you see the limits of the resolution, but it's hiding what's really going on. The color chart should have been entirely black and white. That chart is black and white, as printed on a black and white printer. Anywhere that you see color at all, that's false. That was an artifact introduced by aliasing. That is not an accurate reproduction of the chart. Anywhere you see orange or purple on that chart, that's an artifact coming from the camera's aliasing.

[Chris Barcellos]

Barry:

This chart was printed from an available on line .pdf, the printer was a Brother 9700 Laser. They are 600 dpi x 600 dpi.

Thank you for the information about how to read these. I was also concerned, as stated in my post, as to how good a chart like this could be relied on.

In meantime, I am curious if you think shooting with reduced sharpness and the "superflat" Picture Style or one similar that has posted that provide a flat and unsharpened image that can be enhanced in post can help resolve some of issues. Have you done any experimenting there.

[Barry Green]

The author of superflat now says "don't bother, just use Neutral and set the contrast and sharpness to minimal." I haven't seen that post, so I don't know about the accuracy of it.

I've tested the SLRs with res charts six ways from sunday, and the contrast and detail settings don't really have any effect on the aliasing. I mean, juicing up the contrast can accentuate the visibility of the jaggy lines, but they're still jaggy even at minimal contrast, so it's all still there.

If you printed your chart at normal size, I don't think it's doing you any favors. Try printing it at 4x size and reshooting the chart, you might see cleaner results. I have a "1200x1200 dpi" laser, and ... it's bogus. It can't print a 600dpi zone plate to save its life. So if you want the best results you can get from that chart, try printing it at 4x size.

If you want to overcome the aliasing and remove it from your equation, there's pretty much only one way to do it, and that's to lower the amount of overall detail that's getting passed to the chip. Whether through an on-chip anti-alias filter, or through something like the Caprock Anti-Moire filters, or through just defocusing the lens a little. Aliasing happens when the lens is passing too-sharp image information for the sampling system to properly and accurately represent, so (since you can't put in a better sensor in there) the only solution is to stop the lens from passing too much detail.

The cost of doing so, of course, is that the resulting image will lose all its HD "snap". It'll become a good, solid, clean image, but it'll be more along the lines of a 720p image, because that's about what the DSLRs can cleanly resolve.

[Chris Barcellos]

Thanks, Barry !

Never got into this resolution issue much, but thought it might be smart to learn a little bit about it when the subject comes up, by shooting a chart or two. I will try your recommendation of printing 4x.

[Tom Roper]

Nothing beats a properly designed chart, but it's not the best way to pull a number. For example, Barry is looking for accurately rendered uniform lines, or the point where the camera quits representing the chart accurately, but at what contrast level does it happen at? MTF100? 90? 80? 70? You get a different resolution number for each step in contrast. Two people won't even read the chart and be within 100 numbers. What's the value in that?

When it comes to pulling a number absolute, this is where I think Imatest is the better way to go. The software also normalizes in-camera sharpening to a standard level, usually 2 pixel radius.

Now what I told you Chris, (you didn't seem to listen) when I said you can't compare your numbers or my numbers to someone else's, (which you did) because everybody uses different methods. If you photographed your chart the same way as Barry, and he interprets your chart he should come up with similar numbers to what he already reported, using whatever level of contrast is appropriate to his eye, as long as he is consistent. But you are trying to say his numbers are wrong, somebody else read them higher. And now you have seen that Barry reads your chart more conservatively than I, so you should see the folly of trying to use a number.

For eyeballing a chart, the best way is to shoot the same chart the same way with two cameras, and view the chart with your eyeballs and choose the camera with the more pleasing rendition. That's a valid test.

I also said that the 5DMkII tested lower resolution than the HV10, XH-A1 or EX1, in a side by side comparison using the same Imatest method. But you locked in on the number "800" which was what Imatest reported for the XH-A1, not the 5DMkII. It clearly is ahead of the 5DMkII.

The folly of resolution testing...

[Chris Barcellos]

"Now what I told you Chris, (you didn't seem to listen) when I said you can't compare your numbers or my numbers to someone else's, (which you did) because everybody uses different methods. If you photographed your chart the same way as Barry, and he interprets your chart he should come up with similar numbers to what he already reported, using whatever level of contrast is appropriate to his eye, as long as he is consistent. But you are trying to say his numbers are wrong, somebody else read them higher. And now you have seen that Barry reads your chart more conservatively than I, so you should see the folly of trying to use a number."

As I indicated to Barry in my last post, I am interested in learning about testing. I fully intend to test against my own HV20, and my FX1, as you suggested, so I can get relative values. Again, I am not really worried about numbers I get out of any of them, and I certainly am not going to debate the merits of the 5DII based on my tests. My reference to your post in this thread talks about using the flat settings, and does refer to you post in the other thread, and wondering if others could check it out. I merely stated that at least one person interpreted at 800. But I am curious how the tests themselves are conducted, and if the use of super flat type settings can influenece the end result. It is not my intention to start a pixel counting war. Thanks for your input on this too.

[Ian G. Thompson]

....thanks guys. I've learned something new about using charts.

[Pat Reddy]

Chris, I accused myself of being a pixel peeper, but I'm not sure anyone really called you one here (unless I missed it). My point is that I think it is a mistake to say that the 5D II is not capable of true 1080p. It has resolution and artifacting issues for sure, but none of the cameras we talk about on dvinfo could be said to have true 1080p or 1080i if you are using resolution tests as the key benchmark. That doesn't mean it's inappropriate to test them and to base decisions on those tests.

It's pretty obvious that the 5dII and & 7D are in a class by themselves: they are capable of doing things that hugely expensive cameras can't come close to and at the same time they can't do some basic things as well as the least expensive cameras. I know that in my hands at least the 5DII is capable of really ugly or really stunning video, and I am having to learn how to manage things in ways I never had to worry about with other cameras.

I share your frustration with their limitations and excitement about their capabilities.

Pat