A word on Nikon D5300

I think it is the first Nikon D that has non-zero black level. On the shots we got it is at 600.
Well, it's about time; they've finally come to their senses. Clipping away negative swings of read noise limits the effectiveness of RAW-level noise reduction, and sometimes individual cameras seem to have blotchy bias problems which could be corrected with an unclipped black frame.

Converters, of course, need to maintain negative values until conversion to display RGB, and allow corrective frames for this to have full benefit. Even downsampling should be done with negative swing intact; it improves the linearity of deep shadows.
 
Thanks Iliah. Maybe Nikon learned their lesson with the horrible unbalanced clipping of the D7100/D5200.
 
Thanks Iliah. Do you have any insight to whether this is the same basic sensor or something new? The promotional images of the sensor itself suggest some changes..
 
so it looks like they have very simillar sensors (5200 and 5300)

but some minor differnces are there.

i wonder if it is the same sensor or not.
 
so it looks like they have very simillar sensors (5200 and 5300)
Look at the Nikon web site and see they are completely different.

And of course the raw pre-processing is different.

I do not see DxO mark scores to be useful when comparing high end sensors to each other.
 
Four cameras with four different sensors in 15 months. But looking at the dxomark benchmarks, there isn't much to separate them, at least until you put a lens on. At that point the D7100 stands out, and I expect the D5300 to perform similarly. I'm betting that this one is a modified Toshiba but other than cost savings, I think it is a legitimate question just what Nikon is looking for here.
 
So what methodology would you suggest?
Well, if you are going to start a camera testing site we sure can talk.

--
http://www.libraw.org/
http://www.rawdigger.com/
;)

i was thinking more for personal use...

although i also sometimes review gear so i would be happy to learn.

how can i test the quality of a given sensor given i have the two cameras i wish to compare and am interested in their raw performance which would be more insightful or maybe complimentary to what dxo are doing?

so i know they are not testing banding... and pattern noise

what substantial factors should be of interest?
 
I think it is the first Nikon D that has non-zero black level. On the shots we got it is at 600.
Now that is interesting. It also might explain the slightly higher effective read noise levels reported for the 5300...or maybe my brain is misfiring.
I would think that DxO would do a good job of estimating the read noise with black-clipped RAW files. However, just estimating the black frame noise before clipping isn't enough, as damage is already done. I don't remember the details of how I calculated this, as it was about a decade ago, but I found that when you clip RAW data at mean black, you clip the signal of very weak ones such that signals close to black only have about 60% the SNR that they would have if the data were not clipped.

This is not to say that most converters would even maintain the negative swing of noise before conversion, though; I suspect that many clip the RAW data at mean black before they do anything else.
 
Four cameras with four different sensors in 15 months. But looking at the dxomark benchmarks, there isn't much to separate them, at least until you put a lens on. At that point the D7100 stands out, and I expect the D5300 to perform similarly. I'm betting that this one is a modified Toshiba but other than cost savings, I think it is a legitimate question just what Nikon is looking for here.
I don't think so, I think it's a Sony. Here is the D5200 sensor (from Nikon's site)



img_01.png


The Toshiba sensors have this solid black mask. Here is the D5300 sensor



img_01.png


It's in a rather unusual package, which means that you can't see all the lands at the edges, one of the good ways of identifying a Sony sensor, but they do have some characteristic patterns which tend to identify Sony

I haven't found a head-on pic of the latest Sony APS-C sensors, but here's one of the A58's



ZTECH_SENSOR.jpg


Similar features to note are the unusual grey band with gold lands at the top (similar pattern), the similar blue mask stopping short of the active frame are (common on Sony sensors) and the very similar pattern at the left hand side of the active area, clipped form the D5300 below:



187ecfae9025485abc604ce1eee8be92.jpg.png

This kind of pattern is very characteristic of Sony sensors, going back to the A700, although this is a bit different from what we've seen before, but I think look carefully you can see a similar layout on the A58. The A58 is a 20MP sensor, and I think this is a development of that, not the old 24MP that went into the NEX7.

--

Bob
 

Keyboard shortcuts

Back
Top