Sharpness test between Nikon D800, Panasonic GH2 and Canon 7D

Danielvilliers

Danielvilliers

Veteran
I did this test to see the sharpness of the D800 in video compared to the other two camera I had, a gh2 and Canon 7d. I am a Nikon photographer who bought a Canon 7D about 2 years ago (waiting, waiting, waiting for Nikon) and then a gh2 about 1 year ago (because it was so good at video and the hack). Now I just got my D800 (principally for my photography job) I decided to test it against Canon 7D and a stock and hacked gh2.

This is strictly a sharpness test, so I did not care about exposure. I could have shot with my lighting (flash halogen) but I had time this afternoon and I was not sure if I would have had enough light to close to f8 or more to try to decrease the problem of DOF.


comparison 7d.jpgcomparison canon 7d.jpgcomparison canon d800.jpgcomparison canon gh2 mysteron.jpgcomparison d800.jpgcomparison gh2 mysteron.jpgcomparison gh2.jpgcomparison gh2.jpg


If ever those image don't show up at full 1080p resolution you can see them here

https://picasaweb.google.com/115991...anasonicGh2Gh2Canon7dMatchToAboutSameContrast

I have tried to match the contrast on the files above to have a better sense of sharpness and detail. If you want to see the whole bunch of files including cinestyle etc you look here

https://picasaweb.google.com/115991370373823953401/TestSharpnessPanasonicGh2NikonD800Canon7dInVideo

My observation is that the gh2 is the sharpest, the hack one a little bit more but where it makes a difference is in the detail that it captured. The D800 is a close second, the image is sharp and I needed to flip between them in photoshop to see that the gh2 is sharper. You do see some moire on the red scarf around the dog neck . The 7d is just terrible in terms of sharpness and detail compared to the other two. I clearly took care to focus on all of them with zoom on the lcd.

The point focus being the rose stripe in the middle. It would have been better to have flatter subject/plane and bigger so that the distance is bigger to get into infinity. but I think this test is enough if you are not the type of 400% pixel peeper.

Now my conclusion, is that the d800 could clearly be a very very good video dslr/camera. If the report of the mosaic is as good as it is it would remove the only problem (moire/aliasing) of this camera. This test shows that it is a true 1080p high resolution camera, the DR (will do test shortly) and the rolling shutter also seems to be big improvement compared to the other SLR. Now pare it with a $ 400 blackmagic recorder or $ 1000 Atomos Ninja and we are in another dimension.
 
Last edited:
nope it doesn't but i was looking forward to your comparisons on here Danielvilliers.
The rolling shutter thing...it looks like both 5d3 and d800 are not so good here.....watch .review at 13:10.
 
I find it strange because the links works for me. Can anyone confirm if they can or cannot connect to the picassa web gallery where you can download the files?

Without https:// it is like below, if else does someone has a system for me to post it?

picasaweb.google.com/115991370373823953401/SharpnessTestNikonD800PanasonicGh2Gh2Canon7dMatchToAboutSameContrast#5763976196242465250
 
Danielvilliers said:
I find it strange because the links works for me. Can anyone confirm if they can or cannot connect to the picassa web gallery where you can download the files?

Without https:// it is like below, if else does someone has a system for me to post it?

picasaweb.google.com/115991370373823953401/SharpnessTestNikonD800PanasonicGh2Gh2Canon7dMatchToAboutSameContrast#5763976196242465250
Click to expand...

Nope does not work:

"Sorry, that page was not found."
 
I did a very similar test recently between the D800 and GH2 (flowmotion hack), and found like the OP that the GH2 had superior resolution, Im not certain how this could be improved by fixing the moire/aliasing with an over-the-sensor filter on the D800. It seems that the bottom line is that the $3K ff camera needs significant hardware tweaking and an offboard recording solution to compete with the $700 Panasonic. I really wanted to like the Nikon, the DR is sweet and the build quality is superb, but ultimately I think it is hard to justify unless I really needed FF stills for some reason. In my test I used the same lenses on both cameras, a 24-70 f2.8 G and a Tokina 300mm f2.8 AT-X Pro, all exterior daylight. If anyone is interested PM me for a link to the test footage.
 
LyleGorch said:
I did a very similar test recently between the D800 and GH2 (flowmotion hack), and found like the OP that the GH2 had superior resolution, Im not certain how this could be improved by fixing the moire/aliasing with an over-the-sensor filter on the D800. It seems that the bottom line is that the $3K ff camera needs significant hardware tweaking and an offboard recording solution to compete with the $700 Panasonic. I really wanted to like the Nikon, the DR is sweet and the build quality is superb, but ultimately I think it is hard to justify unless I really needed FF stills for some reason. In my test I used the same lenses on both cameras, a 24-70 f2.8 G and a Tokina 300mm f2.8 AT-X Pro, all exterior daylight. If anyone is interested PM me for a link to the test footage.
Click to expand...

For me, I think that if it is strictly for video the d800 might not be the best investment compared to the hacked gh2 or better still (will have to see test) the Blackmagic camera for $ 3000, unless the user wants the vista vision look. But for me it could still be a better choice than the hacked gh2, because mainly of the DR and from what I feel the jello. I will have to do some test of both during the week, but it seems that it is better, to what extent will dictate my choice to be my A-cam for video.

Personally this test tells me that the D800 is sharp enough for my use. It is a little less sharp than the gh2 but you have to put them side by side to see the difference. The gh2 itself has compared very favorably to the pro S35 1080p camera like the Sony F3, Canon C300. Now we will have to see from test about the mosaic filter and an external recorder how it finally ranks compared to other dslr and camera.
 
I've found very similar results: it's not as sharp as the GH2, but it's definitely a lot sharper than the Canons, which look terribly soft side-by-side. (BTW: the 5N is nearly as sharp as the D800, and has similar DR too; my shootout with these 4 cameras is nearly done, it shoul go online later this week)
 
I'd also love to see versions with sharpening applied in CS6/FCP, etc.

The Canon sensors have always needed some pretty aggressive post-capture sharpening to counteract the strong AA filter. For still images, Canon recommended using an Unsharp Mask set at (300,.1,0) for their 16MP Canon 1DsII. I have found that works well as a guidline for their current 18MP sensors, including T3i video images. I often vary the second parameter up to .9 pixel for non-portrait images.

In general, there are 3 types of sharpening: post capture to counter the AA filter, creative, and output sharpening (calibrated for the output size and device.)

I have also found it very userful to run the same "Auto Color" color correction routine on all captures when comparing images, so that the images look as close as possible. I have found that certain color casts give an impression that an image has more content/sharpness/detail. "Auto Levels" is also useful in the same way.

Looking forward to the tests! Thanks folks.

Best,
Michael
 
You can use this link

https://picasaweb.google.com/115991...oAboutSameContrast?authuser=0&feat=directlink

For just the set that I have more or less normalise so that they look lose so that contrast does not influence sharpness or the one below when you will see a series with some Canon cinestyle one etc... You just can put some sharpness filter on the png one s that are not trully compressed. The d800 ones sharpens very well (Might make some artifact a little bit more apparent).

https://picasaweb.google.com/115991...D800Canon7dInVideo?authuser=0&feat=directlink
 
Here it is: the Flaat Shootout 2012
550D (T2i) vs NEX-5N vs GH2 vs D800
https://vimeo.com/46037571

my computer has some kind of heat-related fever and is creating data corruption at rendering, I'll try to post a fixed version in a few weeks, but I'm not going to have a lot of time for this, so, just in case, there you go :)
 
Hi Samuel, thanks for the test. I repead just what I commented on vimeo below.

I would say that the general consensus is that the flattest neutral profile is Smooth -2 contrast with the gh2. I think that standard - 2 is a bit more contrasty.


The second thing is that you can change the gama with 5dtorgb conversion. When you combine it with hacked gh2 footage it can give much flatter images because the hack preserve a ton of detail in the shadows.


I did a test comparing the hacked gh2 to Canon 7d below.
personal-view.com/talks/discussion/1795/dynamic-range-test-between-gh2-and-7d/p1
 
Yes, I've seen the comments on vimeo (yours and from others) asking for 5DtoRGB conversion of the GH2 footage. I'll do that when I re-upload the footage to fix the data corruption artifacts. It may take me some time to get back to this, though... Cheers!
 
If you filter the GH2 footage with a 5DtoRGB transcode, you'll no longer be evaluating the original GH2 footage. You'll effectively be using a chroma smoothing filter to produce an "enhanced" version of the original.
 
lpowell said:
If you filter the GH2 footage with a 5DtoRGB transcode, you'll no longer be evaluating the original GH2 footage. You'll effectively be using a chroma smoothing filter to produce an "enhanced" version of the original.
Click to expand...

I suppose the implied follow-up to that is:

"What would be the best-in-class way to process the footage from each of the other cameras?"​



I'd like to hear what people think for each camera.

On the Canon, maybe an ISO of 160 with ML? What about the newer 600D and 650D? The other cameras are current, while the 550D is 1 to 2 generations older.

(I know you tested what you had Samuel - Thank You!!) :)

Michael
 
unadog said:
"What would be the best-in-class way to process the footage from each of the other cameras?"​
Click to expand...
That would be the appropriate question if what you were aiming to evaluate were the results of your complete post-production workflow. But in this case, what the OP set out to evaluate was the sharpness of each camera's native footage. For that purpose, the best way to evaluate the footage is with no additional processing or transcoding at all.
 
lpowell said:
For that purpose, the best way to evaluate the footage is with no additional processing or transcoding at all.
Click to expand...

I understand. That was in response to the question about the extra post processing on the GH2.

I think the reality for Samuel is that there are an infinte number of variables to take into consideration. Of course, it is completely up to him which he thinks are worthwhile investigating.

M.
 
I think in the end it is the result that count. If you talk DR, then the hacked gh2 has a lot of detail that are preserved in the shadows because it is not compressed into mush by the encoder because of the bit-rate. Just have a look at the link that I put on the test that I did on my post above. The gama shift (lower contrast) that people are using in 5dtorgb is just a perception of something that is already here, you could achieve the same with a curve lift of the shadows.
 
You must log in or register to reply here.
Back
Top