Nikon D850 medium RAW files soft?
In the review of the Nikon D850 camera, I noted that this near-perfect camera had one major flaw for me – the medium RAW files appear soft. For event work or any kind of volume work, the massive full RAW file is just too much overhead, whether in storage or processing time. I need a RAW file in the 20-24 megapixel range. The medium RAW file of the Nikon D850 (B&H / Amazon) would have been ideal – allowing me to shoot the majority of work as medium RAW, and occasionally flipping over to full-size RAW.
I have had several photographers contact me to say they just don’t see the same results as I had. One of the things that I take pride in with the Tangents blog is that I want everything that I show and explain, to be real and stand up to scrutiny. That’s one of the motives behind this website – with the Flash Photography Tutorials, I wanted to be able to show the results from the techniques, and not just make big claims or grand-stand, as you often see on photography forums or in Facebook groups. I want the material to show in consequent way, what the flash photography techniques can achieve. The same goes for everything throughout the site.
In testing resolution of a lens, I am mostly guided by how sharp a person’s eyelashes are rendered. That helps me decide if something is really sharp … or just adequate. It works for me, but it isn’t exactly a neutral, scientific test.
The idea that I might be propagating a false idea that I found the Nikon D850 medium RAW files to be soft, pushed me to finally buy a lens resolution chart – this one. It is made by AbelCine. More about it on their FAQ. There’s still a learning curve for me in properly interpreting the info to be gleaned from it. But at least, there is now a less subjective way of comparing results. (That thing is expensive, so expect to see it more often here on this site!)
I took photos of this in the studio, using 4 cameras:
- 45 mpx; 25mpx … Nikon D850 (B&H / Amazon)
The example shown below is with the camera set to medium RAW, and processed with ACR. - 20mpx … Nikon D5 (B&H / Amazon)
The example shown here is with the 20 mpx image rezzed up to 25mpx so that the size is the same. - 24 mpx … Nikon D750 (B&H / Amazon)
- 36 mpx as 25mpx … Nikon D810 (B&H / Amazon)
I shot with the D810 in the 1.2x crop mode, which brings the 36 mpx down to a more manageable 25mpx which we can then compare. I had to move the camera slightly back to get the same framing.
All these were shot in the studio using studio lighting, with a Nikon 105mm f/2.8 micro lens set to f/8. That lens at f/8 is really, really sharp. I shot this at a distance of around 12 ft to the board.
Here are 100% crops of approximately the same area of the processed JPGs of the RAW files. I tried to match the framing as closely as possible.
What I looked at – the sharpness of the lettering, as well as the sharpness of the two areas marked 9.5 – the softness might not be dramatic, but it is there.
Now specifically look at the larger area with concentric circles marked as 7.5 – the D850 medium RAW file doesn’t show this as perfectly concentric. There is some weird artifacting happening that will affect detail. The other three cameras don’t show that.
To my eye, the Nikon D750 looked sharper than the Nikon D850 medium RAW file. The 25mpx crop of the Nikon D810 looks sharper than the medium D850 file. The uprezzed Nikon D5 file still looks a touch sharper to me than the medium D850. With the superior focusing of the Nikon D5, and the superb high-ISO noise, this is till my first choice as a workhorse camera.
Summary
There are the results in a more tightly comparable format. Whether the differences are enough to swing you either way, is personal choice. I wanted to be wrong about this – in so many respects the Nikon D850 (B&H / Amazon) is as perfect a camera as you can possibly get. But there it is. As mentioned in my original review of the Nikon D850, it convinced me that I needed another Nikon D5 (B&H / Amazon).
Related articles
- review: Nikon D850 camera
- review: Nikon D850 high ISO test
- more reviews of Nikon gear
You are absolutely correct, I’ve now looked at this for over 5 minutes, taking a break from screen on several occasions, coming back and sliding the different images.
The thing that also stands out to me is if you look at the bottom middle of each image, between the fat short yellow line to the black curved lines next to it, those lines stand out slightly sharper then the 850 does.
That 750 is quite sharp in that area compared to the 850.
So yeah, I concur.
Trev.
Well.. there it is. I’ll be damned, you’re right. Stick with the D5.
Thank you Neil for your work. Just today I have recommended Tangents to a budding photographer I have met.
I hope you or someone would create these comparisons for other pairs or groups of cameras that are typically considered as alternatives or with similar sensors. E.g. D4 vs Df or, in a time-machine of sorts, the D3 vs D700.
With D850, could it be sample variation or some smudge on the sensor?
Marcel … in time I will do this for more cameras and lenses.
Regarding whether this was a sample variation … I only had the one camera to test. That in itself is a limitation. But I am sure this is a problem for all D850 cameras. And no, the sensor is pristine.
At a time of huge buffers and cards, is an oversized s/mRAW format useful much? They don’t seem to save as much space as the loss in res would suggest.
Yes, for any kind of volume work, a massive RAW file is a potential problem.
As one example, with time-lapse photography, I now use the Nikon D810 with a crop-sensor lens – the Nikon 10-24mm – so that I only have to deal with a 15megapixel / 20 Mb file.
What tripped me over to do this, was when I shot 40,000 images for a time-lapse job, and it took nearly two weeks for the 40,000 RAW full-resolution D810 RAW files to process to JPGs. The lens corrections especially slowed it down.
So by now shooting a 15 megapixel file, instead of a 36 megapixel file, I have more than halved my processing time. This becomes crucial in giving a faster turn-around time for a client.
I am sure that there are other examples too where a massive RAW file is not a bonus.
thing is medium raw has to interpolate which will cause moire when not dealing with exact multiples when resampling.
just grabbing the res. specs
(L) 8,256 x 5,504 (45.4 million) (M) 6,192 x 4,128 (25.5 million) (S) 4,128 x 2,752 (11.3 million)
the small size is exactly half the lines in each direction each pixel is the average of a 2×2 square with medium you have to take a 4×4 pizza and recut it into a 3×3, it will be messy.
Down-sizing the full-rez JPG to 25 mpx looks really good though …
in camera or in post?
In post. See the comparison in the original review.
That would explain the difference. You have way more processing time & power in post compared to in camera.
To enable fastest RAW write times speed will be paramount over quality. Otherwise, you’ll need faster processors in the camera and more ram etc. which will eat battery life and add weight/bulk.
Spot on – the D750 sensor is one schweet sensor – wish it was in the D4/ D4s or D5
Except that the Nikon D5 is unbeatable for high-ISO noise performance.
What does Nikon do to Raw files when Medium Raw format is selected (vs large Raw files)? I understand a low resolution JPEG file is achieved via high(er) compression, but Raw files aren’t compressed. Since Raw files are directly out of camera, Nikon must exclude certain pixels on Medium Raw files (vs large Raw files). Anyone know the answer to how lower resolution Raw files are handled in camera?
Eric, for more info on how the Nikon D850 sensor works, and how the medium RAW file is processed, check out this authoritative article on the PhotographyLife website.
I just noticed this, unfortunately, when a feature I shot for a magazine turned out pretty soft in print. I had cut the file size to medium when my editor told me that my files were just too big to handle. Think I will use the D5 for this purpose from now on, or even the 750. Going back over other shoots I have done recently, including portraits and performance/sports–those files look a bit mushy compared to my workhorse Nikons.
Yes, Nikon D5 is a well better alternative! The only exception is for when you need a big fat photo. :)