Mother of God - D800 scores 95 DxOMark
- Thread starter Wrathwilde
- Start date
- Status
B
briansquibb
Guest
ssrdd said:
I can tell you have done your research - you spotted the 'knock the 5DIII and praise the D800' bandwagon and jumped on.
DXOMark comparing the 5DIII and the D800
http://www.dxomark.com/index.php/Cameras/Compare-Camera-Sensors/Compare-cameras-side-by-side/%28appareil1%29/795|0/%28brand%29/Canon/%28appareil2%29/792|0/%28brand2%29/Nikon
Upvote
0
Spriter said:
I'd argue most of us crop the same regardless of mp count since we crop for composition an not resolution. Thus the advantages of low resolution sensors are nullified. Dxo's arguments is that pixel quality 100% crops are not representative of actual quality for most uses and thus they normalize the field by the downscale. If you're selling tiny 100% crops blown to 8x10, you should disregard their findings. But if you sell prints of 90% if your original , their method is more accurate than the side by side comparison of crops you see on the web.
Upvote
0
T
thure1982
Guest
Have you guys matched it with the 5D2?
Sure it got 95 "points" at he scoreboard but the aqual difference is not the obvois when you look at the diagrams.
Sure it got 95 "points" at he scoreboard but the aqual difference is not the obvois when you look at the diagrams.
Upvote
0
briansquibb said:wickidwombat said:i would have read all of this thread but i've been out all weekend shooting with my 5Dmk3
since i don't really follow the lab tests or particularly care about them for that matter, 95 is good is it?
Most good bodies score in the 80's
Personally I feel the score is so much higher that this cynical old man wonders about the accuracy
That would make it about 1.5 stops better than current APS-C bodies, which seems plausible. Full frame bodies aren't released very frequently, so there aren't that many comparable bodies that have been tested. However, Nikon do have a couple of other FF models that score close to 90.
Upvote
0
D
D.Sim
Guest
ssrdd said:
So sayeth the Nikon fanboy...
Gee I miss the karma system... some of the comments are just getting out of hand -_-
Upvote
0
briansquibb said:
As far as DxO's preliminary comparison, look at the specs... Canon 5d mark III = professional. Nikon D800 = Semi-Pro DSLR. Now let me duck out before the bullets/knives/forks come flying out from the left.
Upvote
0
B
briansquibb
Guest
awinphoto said:
DxO haven't done a review of the 5DIII therefore any comment against it must have been made by a Nikon fanboy. Personally I dont care less - I am not getting a 5DIII as I already have a camera that fills that requirement.
Upvote
0
Um. Isn't that the D4 (well, 16MP instead of 22MP, so even more advantage)? And yet DXO rates the D4 as having less DR than the D800? Normalized or not, that makes me think:
On a per-pixel level, SNR is absolutely related to pixel size. Upon normalization, in this case downsampling (or pixel binning), the difference might become negligible between a higher resolution sensor & a lower resolution one. But, in my understanding, the DR or SNR can never be better for a downsized image from a higher resolution sensor vs. the image from the lower resolution sensor w/ bigger pixels.
Why?
Because the higher resolution sensor will have more read noise because it had to read many more pixels. Yes, the read noise will be averaged upon downsizing, but averaging the read noise of, say, 4 pixels, down to one can't possibly give you lower noise than reading just 1 pixel to begin with off a sensor of 1/4 the resolution.
Hardware-binning, on the other hand, was invented to combat this problem. Phase One's Sensor+ tech bins pixels at the sensor level so that you don't have as many read events when you shoot at a lower resolution (on the same sensor), thereby increasing your overall SNR.
Until the greater inter-pixel spacing of higher resolution cameras no longer affect QE, & read noise goes to 0 (not likely), high resolution sensors cannot have the same SNR performance of lower resolution sensors of the same generation, barring other factors that might make the higher resolution sensor better.
This is likely because of dirty downstream signal processing, which Nikon/Sony has less of. Therefore, the signal at which SNR = 1 for Nikon/Sony images is lower than for Canon; hence the higher DR at lower ISO. ISO 400 is unity gain for 5DII; below that ISO, >1 electron is used per 1 digital unit (DU) in the resulting image file. For example, at ISO 100, 4e- = 1DU (I believe), which means that lower signals are now more prone to noise injection from dirty electronics.
That's my understanding, anyway. Feel free to discuss.
Also, though't I'd point out that according to sensorgen.info, which basically pulls data from DXOMark, the D7000's read noise remains the same even at low ISO, something pretty much atypical of all other cameras:
http://sensorgen.info/NikonD7000.html
This also has the effect of making the DR on the D7000 literally plummet as soon as you raise the ISO. By ISO 200, it's down to D4 levels of DR at ISO 100; The D4 DR drops rather slowly from ISO 100 to ISO 800 (http://sensorgen.info/NikonD4.html).
But the D7000 having a lower read noise at ISO 100 than the D4 and higher DR than the D4 by 1.4 stops?
Really??
Upvote
0
if DXO wants to do a FAIR test on any sensor, they should use the SAME optics, SAME shutter, SAME aperture assembly, SAME wave length of light, and SAME all other things to rule out any other factors. From cnet review I see they use canon lens to test canon body, I'd like to think they use nikon lens to test nikon body, etc. That makes the result less legitimate.V8Beast said:
The other way I see is the Full Sensor Apparent Image Quality (FSAIQ) is not the actual image quality, sharpness and contrast plays important roll here, which related to both sensor and lens. So again DXO does not tell the sensor itself, it does tell the appearance in a specific case (8MP).
Somewhere at clarkvision website says the optimal size of a pixel is 5um and the density of a FF is around 34MP as D800.
I just want to be a little less bias.
Upvote
0
Not necessarily. You can take multiple exposures at different exposures, &, for example, select the RAW file for each body that is just short of clipping the brightest patch of a wedge (in other words, 1/3EV more exposure clips, say, the green channel). You then find the darkest patch with SNR = 1 (or whatever your criterion is), & as long as you know how many stops are between that dark patch & the bright patch, you can calculate a dynamic range. Any differences in lens transmission & such are compensated for by you selecting the appropriate exposure RAW file for each body.
You can even do this 'in the real world', & if you do, you'll see that a Nikon D7000, D800, or any other of their bodies with the stellar Sony EXMOR sensors will have much cleaner shadows than any Canon 5D series body (or likely any Canon body, period... I'm speaking of the 5D series b/c that's what I own & have tested). Assuming, of course, that your scene has enough dynamic range to begin with (easily demonstrable with sunrise/sunset shots shooting toward the sun).
If you're suggesting that DXO is biased, what evidence do you have of that?
Upvote
0
sarangiman said:
This is common knowledge! a) we all know the 5d3 has a great sensor, b) dxo failed to praise it and rated d800 higher, a+b) dxo is biased, really simple if you think about it :-o
Upvote
0
I don't say that's biased. However, people should too focus on the score.Marsu42 said:
Many benchmark software for CPUs and GPUs and they all have the different results for the performances. Some like nVidia, some like AMD, and some lie Intel. You need to understand how they give the score to understand the meaning of the score.
I am not go that far yet, so the DXO score is not important for me. I guess some people do a lot of photo retouching might care about the DXO score. my 2cents.
Upvote
0
After working professionally now with my 5D3 I have come to believe the D800 probably is a better stills camera (though I haven't shot with it) because of that sensor. But the 5D3 is a better video camera (even though resolution crippled). Event video with the 5D3 is outstanding (ISO 5000 is awesome). The RAW stills on the 5D3 are excellent but can't be pushed as hard in post as I've seen done with the D800.
Still, it would cost me over $10K to transition to D800 now, and the video wouldn't be as good. And I imagine Canon will have to respond within a year or so with an EF camera that can compete for stills. For video, Canon is saying C300 or forget it at this time. Invest in lenses instead at this point (next on my wish list is the 17 TS-E of course).
Still, it would cost me over $10K to transition to D800 now, and the video wouldn't be as good. And I imagine Canon will have to respond within a year or so with an EF camera that can compete for stills. For video, Canon is saying C300 or forget it at this time. Invest in lenses instead at this point (next on my wish list is the 17 TS-E of course).
Upvote
0
That's what many people argue here. Many people here are shooting for events which is usually in low light environment. D800 cannot bring much benefits for them even though it has more details and better DR.peederj said:
I am just take photos for my family and for fun, so low ISO and high ISO is about 60/40. That's why D800 is very attractive for me. Hopefully Canon will have an FF camera similar to D800.
Upvote
0
I'm also currently shooting with the 5D Mark III; its AF performance w/ my f/1.2 & f/1.4 primes is, so far, what I'd always wanted/expected in a high end camera.
But that doesn't mean I don't envy those game-changing sensors in the Nikons, especially for landscapes & environmental portraiture.
But the D800 doesn't really hurt either. When the D800 image is downsized to 5D Mark III sizes, it's ISO 6400 images are virtually identical to the 5D Mark III image in terms of noise. I haven't compared higher ISOs myself. Video at higher ISOs do suffer on the D800 though.
But that doesn't mean I don't envy those game-changing sensors in the Nikons, especially for landscapes & environmental portraiture.
But the D800 doesn't really hurt either. When the D800 image is downsized to 5D Mark III sizes, it's ISO 6400 images are virtually identical to the 5D Mark III image in terms of noise. I haven't compared higher ISOs myself. Video at higher ISOs do suffer on the D800 though.
Upvote
0
J
Jotho
Guest
wickidwombat said:i would have read all of this thread but i've been out all weekend shooting with my 5Dmk3
since i don't really follow the lab tests or particularly care about them for that matter, 95 is good is it?
Second to that. There will always be something better out there. I can't say if D800 is better because I. don't care. I've had my MkIII a couple of weeks now and it's just great. Being just an ethusiast I have learnt tons already. Yesterday even my wife commented that the pics looks greater than with my old 60D. I'll continue shoot this weekend and not worry too much about what Nikon does, I hope all D800 owners are just as happy I am though.
Upvote
0
Agree. However, that only works for people like me, not a PRO. I don't think people shoot 500+ photos for an wedding could spend time to downsize those photos and showing to their customers. That's too much effort.sarangiman said:
Upvote
0
sarangiman said:
OK, if, you select the "darkest patch with SNR = 1", that is the base noise (or white noise if in video), then, "as long as you know how many stops are between that dark patch & the bright patch" which I think is your "just short of clipping the brightest patch of a wedge", that is your complete white, the final result is the maximum stop you can get, not a problem.
The problem is, since the complete white is all the same in every sensor with same bit length of AtoD converter, you are actually compare the base noise here. You have to assume all the sensors have same overall gain (sensor QE, opamp gain, A/D depth, etc) in order to say this noise is sensor related ONLY, however, since there are many electronic components needed before the GPU(Graphical Process Unit), the result becomes a system overall performance.
The second problem is your "RAW" data is the one AFTER the GPU, which has been processed. It has been well known in the astrophotography that Nikon does NOT output real raw data, some degrees of NR is applied to their RAW. You need to use mode3 to get real RAW in early years, so that "RAW" is not reliable/trustable/usable/representable to the system/sensor.
The third problem, which you did not understand from my post, is the lens DOES affect the result. Think about putting a diffusion filter before an ideal lens, it will mess the very dark and very bright. Let's put another grey color filter, there is no maximum white to be represented by the sensor any more. In theory, the final result is the transfer function of the lens+sensor+converter+processing. I just don't see how you can say the DXO score IS the sensor only properties.
I think the best part of Nikon SYSTEM is the DR curve at low light, which in real life is how much you can pull out from shadow.
As I said, Nikon+Sony, picks a good MP size, have good low noise, processes the data in a optimal way. BUT, this does not mean Canon does not have any of these currently, and "far behind".
I just don't like Canon do not do it.
Upvote
0
- Status
Similar threads
- Locked
- Contains 1 staff post(s)
