5D3 vs D800 - A Technical Analysis (UPDATED & EXTENDED) - With Samples

Started Mar 30, 2012 | Discussions thread
OP Legion5 Senior Member • Posts: 1,047
Replies to Various Posts

SubPrime wrote:

Legion5 wrote:

I didn't see your first post, sorry about that, I can't read every reply. I'm not sure which specific values you're discussing but I used around 30 sharpening and 60 noise reduction for the high ISO tests. Let me know if you have a more specific request.

Thank you but you could have opened the files again in ACR and read off the values you used when you last opened them - they are stored in the XMP sidecar file.

I also asked you the following:
1. Did you use the same values for both raw files?
2. Did you resize using ACR or in Photoshop?

3. Why didn't you merge the 2 into one file without NR and shaprening and then apply NR and sharpening then?

Send me a PM with your e-mail and I'll e-mail you the XMP files if you want to play around with them.

All told nearly every contrast and color parameter was adjusted to correct for color cast and contrast differences between the two cameras as much as possible so it's not just two values.

1. No the 5D3 required less noise reduction to reach an identical level of grain, also because the files are of a different resolution values where pixel size was entered (ie sharpening) had to be equivalent not necessarily equal.

2. ACR

3. Noise reduction algorithms didn't understand interpolated noise well, it causes them to not work very well. Ultimately doing everything ACR was the most scientific solution after testing various methods of doing the comparison.

Jan Madsen wrote:

5tve wrote:

Did he really mean 1/2 x focal length hand held ?

He must have ment 2x, but these posts are filled with contradictions/omissions to support some pre-determined view (at all costs).
--

  • Jan

I'm not really sure where you're coming from with that. There is a strong focus on eliminating any bias here. The 1 / 2x focal length is information from a Nikon Fan - a review of the D800 by one of the first Nikon professionals to get their hands on one who noted that you needed a higher than usual shutter speed to take advantage of the extra resolution the D800 has to offer. In the review it was directly noted that the D800 needed twice the shutter speed of it's predecessor after extensive testing, which makes sense from a physics standpoint given it has nearly double the linear resolution, so the distance the camera can move to maintain a sharp image is halved.

The point of this thread is to help people gain understanding. Bias is directly opposed to that. I'd appreciate it if the sensationalist comments were kept to a minimum.

5tve wrote:

With regard to the cnet link, several photographers who have discussed their article feel the cnet 6400 ISO example is a very clear illustration of the detail advantage of the 5D3 at 6400 ISO, certainly the 5D3 is noticeably sharper and the comparison shows that. You're right that article's writer actually feels they are similar enough at that ISO and are different enough right after. Either way what's significant to some isn't to others, that's why both this thread and the article has samples. I'll have to add that correction given the opportunity.

With regard to the dynamic range, I did not compare the D800 in DX mode, while the DX mode chart is similar in it's findings to my comparison, I actually just compared them in real world back to back testing, which is necessary for dynamic range comparisons. Dynamic range data tends to be all over the place, for example DXOMark's dynamic range data is significantly different from what you posted, and differs significantly from the real world. This is because noise tends to exhibit many unusual behaviors towards the extremes of contrast that is poorly represented in tests but very obvious to the naked eye. For example with your data it shows the 5D2 and 5D3 being nearly identical, despite the fact that it's been found there is a significant difference between the 5D2 and 5D3 in the real world with DR. Overall I'd say that the data you posted correlates to a degree with the real world, and DXO's data correlates even better with the real world (strictly speaking about dynamic range). Either way the 5D3 is slightly better at high ISO vs to D800 to varying degrees depending on who you ask by around a quarter stop, all sources point to the advantage being fairly minor so at least the data is close enough.

With regard to the lensrentals test and their conclusions about detail and ISO. I don't think you have an understanding of what their test means. Their test was done without noise reduction on purpose, if you shoot without noise reduction the conclusion that the D800 has a 10% resolution advantage over the 5D3 with the best primes at ISO 3200 holds true, this was brought up as a counter-point in the last thread with examples. However nobody shoots without NR so the point is moot. A lot of tests under lab conditions can be misleading to varying degrees, That's why I made this thread and developed a comparison methodology that was as helpful and realistic as I could make it.

With regard to not sharpening my samples as much as possible, that was a request in the previous thread, where I did so to show the differences better. The more you sharpen the images the more the advantages in each category become apparent, but nobody shoots at 500% sharpening, so this can mislead people into thinking differences are greater than they actually are. The only place I used extreme sharpening was when I was showing two things were identical where maximizing the difference drives home the point.

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow