5D3 Dynamic Range

[Tcapp]

Now that we have beaten lighting to death, it all comes back to the point of, if you absolutely need the DR and it makes financial sense your path is clear, wither add a d800 to your bag next to your canon gear or have a yard sale and switch. If more DR is not an absolute necessity then, you know, just a nice thing to have just in case, then you really don't need to be switching or adding to the bag.

And you know, if the need for it does arise, more times than not it would be on a job and if you knew it was an important job and want that little extra DR, why not go with a rental on body and lens?

Finally, common sense returns!

 

[TrumpetPower!]

And you know, if the need for it does arise, more times than not it would be on a job and if you knew it was an important job and want that little extra DR, why not go with a rental on body and lens?

If the 5DIII isn't going to have enough dynamic range or resolution for your gig, you'd be a fool to rent (or buy) the D800. The improvement you get with the D800 is just too minimal to have a significant real-world impact in all but the most meaningless of marginal instances.

No, if the 5DIII can't cut the mustard, nothing in the 135 format ever will. What you need for those jobs is medium format -- and not the entry-level $10K kit, either. Plan on at least something in the range of the median annual household income.

The good news is that, if you're actually doing gigs where you really, truly need more than the 5DIII can deliver, then you're charging your clients enough that the MF gear will almost come out of your petty cash account. A single body and its lenses certainly won't be the most expensive part of your capital budget, regardless.

Cheers,

b&

 

[LetTheRightLensIn]

And you know, if the need for it does arise, more times than not it would be on a job and if you knew it was an important job and want that little extra DR, why not go with a rental on body and lens?

If the 5DIII isn't going to have enough dynamic range or resolution for your gig, you'd be a fool to rent (or buy) the D800. The improvement you get with the D800 is just too minimal to have a significant real-world impact in all but the most meaningless of marginal instances.

No, if the 5DIII can't cut the mustard, nothing in the 135 format ever will. What you need for those jobs is medium format -- and not the entry-level $10K kit, either. Plan on at least something in the range of the median annual household income.

The good news is that, if you're actually doing gigs where you really, truly need more than the 5DIII can deliver, then you're charging your clients enough that the MF gear will almost come out of your petty cash account. A single body and its lenses certainly won't be the most expensive part of your capital budget, regardless.

Cheers,

b&

Funny how when canon was 1-2 stops better than nikon sensors the nikon sensors were trash and canons sensors made a huge difference and now that canon is THREE stops worse in one area, that even in that area the difference is non-existant other than in the most meaningless of marginal circumstances. LOL

I'm not saying it is the end of the world, but come on with the ridiculous talk. Just admit it is 3 stops worse and that for some things that can be a difference and how much that matters depends upon the person and let's be done with it.

 

[LetTheRightLensIn]

Increasing the DR range of the recorded image does lose data for a scene that does not contain that large of a dynamic range. This is just a mathematical fact--anyone know of the pigeonhole principle? You can't have a RAW file that contains 14 stops of DR and contain as much information in each stop as a RAW file which contains 12 stops of DR. Both RAW files contain 14 bits per color channel, and you can't store those extra two stops without losing data somewhere. The data is lost because the variation between slightly larger changes in color or intensity is "rounded off" to the same value in order to achieve higher DR.

The sensors have linear capture, RAW files are not tone-mapped but stored in linear fashion, they all have enough bits in the format, all the sensors with better DR do is make the shadow have less noise.

The Canon sensors themselves aren't really any worse they have same DR as Sony it is how they get the read from them where it differs so the only difference those high DR ones provide is cleaner shadows.

* The last sentence is completely unbelievably painful to even read:
"Most cameras are noise-limited, not quantizer level limited. This means that once the signal reach the ADC, there is (at most) 14 bits of information from the saturation level and down to the noise floor."

That sentence is equivalent to worshiping RAW and saying that 14 bits is the end-all, be-all of everything--the most data that can possibly be contained in an image. If that were true, it wouldn't even be possible to change the ISO level on the camera, because the camera would be recording everything that could be recorded.

With the Exmor sensors that pretty much is the case. They are basically ISO-less. Shooting ISO100 and raising it in post 4 stops is basically the same result as shooting at ISO1600. With Canon cameras that is not at all the case.

Likewise, a high dynamic range does not comes without trade-offs, unless a higher quality image format is introduced to store the additional stops of data. Personally, I am advocate of a 48-bit color (3x 16-bit RGB) + 16-bit logarithmic luminosity channel, like Sony's new RGB+W encoding, except with more bits to fit into the current processing standard of 64-bit "words" (i.e., chunks of data).
We can think of these encodings all day long, but no one has control of the market, and no one knows what will be successful.

Most cameras can't even handle a basic 14bit linear file so we are not that sort of worry yet.

 

[LetTheRightLensIn]

If your camera captures the scene perfectly without spending an hour rigging things up it's not poor technique.
It's called not wasting time and getting a more natural looking results to boot. And it's called also being able to make more spontaneously shot stuff and large-scale stuff look better.

Actually, flash is a lot like makeup. A good lighting job will look much more natural than natural light, just like a model with a good makeup job doesn't look like she's wearing any makeup at all.

And that really cuts to the heart of the matter, and to why I keep pounding on the fact that, if the 5DIII is inadequate, then so is the D800.

No the D800 might be adequate if the difference is within three stops.

If the scene is so contrasty that you really need 14 clean stops instead of 12 clean stops, you're shooting in bad light. Not insufficient light, but bad light.

And what about when Canon did 10 stops why isn't the new 12 stops they can do bad light? Because Canon can do it??

And the purpose of flash or other modifiers at that point is only secondarily to add light to the scene. That's incidental, an oh-by-the-way benefit. the real purpose is to fix the light. You know? Add depth and dimesion, sculpt the subject, separate it from its background or surroundings, that sort of thing. And I don't give a damn how much you play with sliders in Lightroom or even with a Wacom airbrush, that's stuff you simply can't do in post if you're even coming close to bumping up against the 5DIII's DR limits.

It depends what you are shooting. If it spontaneous and dynamic, careful light sculpting won't work. If it is a grand scene of nature, forget it.

Once more, with feeling: if the 5DIII hase inadequate dynamic range (or megapixels), the answer isn't to be found in the D800. It's to be found in fixing the light or using some other technique (like HDR or graduated ND filters or whatever). And if you need more megapickles, you either need a multi-shot panorama or you need a larger sensor format.

Really so if three stops can make a difference for some scene it doesn't actually make the difference?

Really and before the 1Ds3 and 5D2 I guess you told people who wanted 21MP that the answer was not in 35mm but to go MF, and as soon as Canon has a 40MP camera then you will ok 35mm is fine for 40MP after all?

Really, people. The differences between the two cameras in terms of image quality amounts to little more than a rounding error.

Cheers,

b&

I don't wanna bash the 5D3, but that is ridiculous. 3 stops is hardly a rounding error. In that case the 6fps of the 5D3 in FF compared to 4fps in FF for the D800 is just a mere rounding error then too no? If you 4fps is not enough then 6fps will hardly be the answer either right?? yeah....

 

[Aglet]

Not really. It seems DR is the only thing that matters.

Speaking of DR, I forgot to mention the D5100. With 13.6 stops of DR for $700, you get:

$51 per stop of DR!

OK, now I need to go start some 5DIII vs. D5100 threads so the Canon bashing can persist

Atta boy.
I'll already provided the fodder for that one.

 

[Aglet]

Right now the 14-bit data per color channel in the RAW is scaled appropriately to store the dynamic range of the camera's sensor, whatever that might be. If the scaling was changed, then the image from the camera's sensor could actually contain much more data.

Yes, I realize that ADCs are typically matched (loosely) to the DR of the sensor, considering full-well capacity & noise. And what I'm saying is that for the current sensors in question, your statement that the sensor with lower DR will contain more information for a lower DR scene (that is still within the DR of the camera), if of course the end points of the lower DR sensor were still mapped to the endpoints of the ADC, is not correct because of the read noise of the system. If the read noise were lower (say 1 electrons instead of >20 electrons), then your statement would be correct.

But as of ~2008, Martinec convincingly shows quantization error is largely absent because the read noise is more than adequately sampled (~6 ADU for the 5D Mark II/III!). Therefore, any more 'accurate' representation of the signal, as you are suggesting, will only more accurately represent the noise (fluctuations) within that signal, without any tangible benefit in actual image data.

I think what helpful means is that IF the DR of the ADC was MATCHED to the DR of the scene, then more info could be had from a low DR scene, even with a lower resolution ADC, but he's not describing his idea clearly.

You could certainly get more info from a low DR scene and low DR ADC if they were matched up to the same low and high levels - but there aren't and they won't be. That kind of fiddling is ridiculous to do in a practical ADC system that's used in imaging this way. Not saying it can't be done, it just isn't at this point and doing so would require that a lot more post-processing scaling would have to be done for each image and then you would still have to move it up or down to place the relative data you acquired into an absolute frame of reference. That's why cameras work the way they do now.

Full scale of the ADC needs to be set to the full-well capacity of the sensor, everything else below scaled normally until base noise is a problem. We don't want to putz around by setting ADC's max to 2/3 of full well for this shot because nothing in the scene is that bright. If you were going to do that you'd do better scaling the analog signal from the sensor to match its peak value to ADC max. but that's still not a good system because we don't want relative data, we need absolute data to simplify creating the final image.

14 bits of ADC is good on SoNikon's latest sensors at low ISO
14 bits of ADC is 2 or more LSBits of noise on Canon's, even at low ISO

16 bits might be done on some specialty or analytical cameras with cooled sensors to get super low read noise.

compress all that down with the non-linear gamma we use and we get a useful 8 bit jpg sort of image.

Helpful, if you and the rest of the academics really want to improve SNR and DR performance, you'd devise a non-linear ADC and processing system to better utilize those bits by spreading them over a log function instead of a linear one. Then a 14 bit log ADC could = 14 stops of DR.

 

[sarangiman]

Helpful, if you and the rest of the academics really want to improve SNR and DR performance, you'd devise a non-linear ADC and processing system to better utilize those bits by spreading them over a log function instead of a linear one. Then a 14 bit log ADC could = 14 stops of DR.

But even if you, say, threw a 16-bit ADC into a 5D Mark III, that wouldn't help at all even for low signals, b/c those signals are subject to 6ADU (for a 14-bit ADC) of fluctuation anyway b/c of the ~33 electrons of read noise, correct? All you'd do is represent the noise even more adequately... the read noise would effectively go up to ~31ADU...

So what you're suggesting would only really help for a sensor/read combo that actually has 14 or more stops of DR (I think if read noise were only 1ADU, you might benefit from a little more precision? i.e. read noise of 2 ADU?). Yes?

 

[pete vella]

http://home.comcast.net/~NikonD70/Charts/RN_ADU.htm#EOS%201D%20Mark%20IV_14
Read Noise in ADUs Chart (2012-04-23)

Enjoy ;D

 

[weixing]

Hi,
Just wonder is it the way how Canon record the RAW images that cause this? For example, the dark frame RAW files from Canon 60D include the bias signal data (no light = 2048), while Nikon D700 RAW file remove the bias signal data (no light = 0).

Not sure about how this will affect real life image, but may be this is why Canon images look noisier in the dark shadow than Nikon?

Just my $0.02.

Have a nice day.

 

[sarangiman]

Canon does preserve the noise information more than Nikon does because of this offset, but I haven't seen any data yet that quantitates how much the noise is 'crushed' by this pseudo-noise removal in Nikon cameras.

I'd like to see what the deviation from the expected logarithmic curve one sees in the actual signals of patches of Stouffer Transmission wedge shots with these cameras. That might give some clue as to what degree shadows are crushed in Nikon RAW files, with the added benefit of a seemingly higher SNR in the dark patches due to the effective lowering of noise (which would give them higher DR estimates if you calculate DR by determining the number of stops between the brightest not-blown patch & the darkest patch that still yields SNR=1). I'm attempting to do this now with transmission wedge shots from the 5DIII & the D7000. As of now, the D7000 yields SNR>1 even at patch 42 (13.2EV), but I'm curious if shadow crushing due to Nikon's special processing of low signals ends up not representing the actual gradation between the darkest patches as well as it might otherwise.

I have no doubt that the Nikon/Sony sensors & signal processing yield better DR than Canon (Fred Miranda's comparison is rather convincing!)... I'm just curious if it's as great a difference in the real world as what the DXO testing methodology indicates.

 

[psolberg]

I have a quick question about the dynamic range of the 5D3 which from what I have been reading is behind the D800. I have never owned a nikon and currently own two 5D3s. I understand what dynamic range is and how it is important for retaining details in the highlights and shadows of an image. I guess I would like to know from one who has seen the differences in the two cameras is what am I missing out on? How does a really high DR (better than the 5D3) help unless I am mostly shooting in high contrast lighting situations or am trying to push or pull and image by more than a couple of stops. I have always been really happy with the DR of my 5D2 and now my 5D3 and being a wedding photographer I shoot in about every lighting condition possible. I always shoot JPEG with highlight tone priority enabled and do my best to get the exposure and WB spot on. I also shoot in Faithful picture mode with the contrast turned down one notch. I end up with pretty flat images out of the camera with plenty of details in the highlights and shadows. Unless I really mis the exposure I have never been unhappy with the DR. I almost always end up adding contrast to the picture because there is too much DR and the image looks too flat. I probably won't own a Nikon so I am just curious from those who have seen the difference hands on how big of a difference is it and in what situations will it really be beneficial. It seems like low ISO high DR performance has become more important than high ISO low noise performance. I am not trying to start another debate over the two cameras I just want to know how much better it is and how much of a difference it would really make.

Unlike a lot of people commenting, I have taken the plunge, joined the dark side or whatever they call it, and actually OWN and SHOT with a D800 out in the field under tricky high contrast as well as even smooth controlled scenarios. My experience goes beyond reading reviews or articles about analog/digital SNR and all that stuff. Charts and equations are neat, but when you're in the field, you don't pull a calculator and start thinking about bits and crap like that. You shoot.

Okay, now that you know my angle, I'll make it simple. Whatever many stops of DR the D800 has, it is just destroys any canon DSLR I've ever shot with, 5DII, mKIII whatever. In paticular, at ISO100 - 800. As for the benefits of that range, you ask a very relevant question. Where are the benefits? First of all let's understand who the D800 is for: Landscape, Studio, and Wedding/Portraits. The landscape guys need no explanation. They are always bracketing, using GND filters and pulling all sorts of tricks to get more DR. At $3000 the D800 (or slightly pricier D800E) is a deal since no current 35mm camera can match it . As Lloyd Chambers puts it

I’m not inclined to disagree here. BTW, the rumor I’m hearing indirectly as word on the street from various dealers is that the Canon to Nikon switch is of tidal proportions, unprecedented.

http://diglloyd.com/blog/2012/20120504_3-ReaderComment-5DM3.html

Studio. Resolution aside (which isn't the subject of this thread), I think they benefit less from the broad DR. But many studio photographers need shadows clean of noise for editing purposes and will do complex light setups so the added headroom isn't bad to have. I shoot under sutdio lights rarely and the times I do, I have kept everything within the range of lesser cameras.

Wedding/Portrait.
I just shot an engagement session in NYC and I'm absolutely in love with the D800, and a big reason is it's DR. The highlights slider in LR4 pulls so much detail out of the images that I almost fell of my chair. I was literally pulling sky detail from images where the 5DII would be dead white blown out. I could have done the same with a big strobe but in NYC you can't set those up anywhere. That is by far my favorite thing about this camera.
In another shot, under shade, I incorrectly metered the background so my subject was nearly gone. No problem, I just re-meterd it it. But in post I went to the bad shot and rescued it. Shot at ISO100, I pulled all the detail I wanted out of the darks with nearly no hint of noise and whatever noise there was left, slight LR4 touches erradicated it. Awesome. Knowing I can pull that much out of the shadows and highlights I can now worry less about DR and more about composition and posing and more importantly, can now take shots I couldn't have before without strobes. I'm still going to use strobes naturally since even the D800 won't show all 20 stops of DR in a scene but it has definitively changed my mind about when I need strobes.

Lastly, girls in white dresses and guys in black moving around outdoors can stress your camera's metering system. The D800 has a very sophisticated 91K RGB meter (on par with the 1DX/D4), but even it goofs up sometime because its algorithms react as you expect: as an engineer not an artist. Having to compensate the meter up and down based on situations is extremelly annoying and a daily routine with lesser cameras. The D800's broader DR makes the inevitable glitch (camera or photographer) much less of an issue, if at all. It's one less crap to think about and if you shoot weddings, you already have enough in your mind.

Well hopefuly that sumarizes my own personal experience with the D800 coming from the red team. As usual, your subjects, shooting style and preferences may result in different reactions. So as they say, your mileage may vary. IMO this is the 5DmkIII I hoped for, it's just made by Nikon ;D. sorry.

 

[V8Beast]

Unlike a lot of people commenting, I have taken the plunge, joined the dark side or whatever they call it, and actually OWN and SHOT with a D800 out in the field under tricky high contrast as well as even smooth controlled scenarios. My experience goes beyond reading reviews or articles about analog/digital SNR and all that stuff. Charts and equations are neat, but when you're in the field, you don't pull a calculator and start thinking about bits and crap like that. You shoot.

Amen to that, brudda

I just shot an engagement session in NYC and I'm absolutely in love with the D800, and a big reason is it's DR. The highlights slider in LR4 pulls so much detail out of the images that I almost fell of my chair. I was literally pulling sky detail from images where the 5DII would be dead white blown out. I could have done the same with a big strobe but in NYC you can't set those up anywhere. That is by far my favorite thing about this camera.
In another shot, under shade, I incorrectly metered the background so my subject was nearly gone. No problem, I just re-meterd it it. But in post I went to the bad shot and rescued it. Shot at ISO100, I pulled all the detail I wanted out of the darks with nearly no hint of noise and whatever noise there was left, slight LR4 touches erradicated it. Awesome. Knowing I can pull that much out of the shadows and highlights I can now worry less about DR and more about composition and posing and more importantly, can now take shots I couldn't have before without strobes. I'm still going to use strobes naturally since even the D800 won't show all 20 stops of DR in a scene but it has definitively changed my mind about when I need strobes.

Lastly, girls in white dresses and guys in black moving around outdoors can stress your camera's metering system. The D800 has a very sophisticated 91K RGB meter (on par with the 1DX/D4), but even it goofs up sometime because its algorithms react as you expect: as an engineer not an artist. Having to compensate the meter up and down based on situations is extremelly annoying and a daily routine with lesser cameras. The D800's broader DR makes the inevitable glitch (camera or photographer) much less of an issue, if at all. It's one less crap to think about and if you shoot weddings, you already have enough in your mind.

Thanks for posting your impressions. I find feedback from the field much more useful than debating lab tests. Looks like you'll be putting your D800's DR to good use, that is whenever Nikon decides to ship it I've been curious why most of the praise has been heaped at the D800's shadow recovery, and not the highlights, but your testing seems to confirm that its DR is great for recovering highlights as well.

Well hopefuly that sumarizes my own personal experience with the D800 coming from the red team. As usual, your subjects, shooting style and preferences may result in different reactions. So as they say, your mileage may vary. IMO this is the 5DmkIII I hoped for, it's just made by Nikon ;D. sorry.

That's funny, because the 5DIII is more of what I expected from the D700 replacement, which is why I own a 5DIII. That makes me wonder how content D700 users are with the D800? Instead of a baby D4 like they were hoping for, they got a D800 with triple the resolution and a slower burst rate than the camera it replaces. That gives them the option of switching from the "low-light, high speed" religion to the "slow-speed, high-resolution" religion, or buying a D4 or a used D3s. You'd think that the D4/D3s option - or switching to Canon for the 5DIII - would be the most practical solution, but switching religions for the sake of fanboyism isn't all that uncommon ;D

 

[well_dunno]

Sorry if this is a noob question but most of what I read talk of photosite size positively correlating with the DR. Somehow the cameras with better DR are not the ones that have the biggest photosites though. I understand SNR is a limiter however not sure if that is the reason for, let us say, mk3 with bigger photosites having worse DR than D800.

I took a look at the thread but could not get it clear for myself. Could anyone shed light on this? :-[

Cheers!

 

[psolberg]

That's funny, because the 5DIII is more of what I expected from the D700 replacement, which is why I own a 5DIII. That makes me wonder how content D700 users are with the D800? Instead of a baby D4 like they were hoping for, they got a D800 with triple the resolution and a slower burst rate than the camera it replaces. That gives them the option of switching from the "low-light, high speed" religion to the "slow-speed, high-resolution" religion, or buying a D4 or a used D3s. You'd think that the D4/D3s option - or switching to Canon for the 5DIII - would be the most practical solution, but switching religions for the sake of fanboyism isn't all that uncom

You make a good point. However, I think that the Nikon crowd remains very well served with the D700 which shoots up to 8fps, has a 51pt AF system of the prior D3 flagship, full frame, and high ISO capabilities between the mkII and mkIII. Lacking video yes, but that aside, for $2200 dollars, it is very good value for a budget fast shooter that isn't concerned with movies. Certainly many canonites wished they had a similar camera from canon for that price. To date, nothing quite matches the full frame D700 speed at that relatively cheap price point. They can buy nearly 3 cameras for the price of one 1DX/D4 flagship body and only give up 4 - 6 MP and a few fps in the process.

So I think Nikonians are a little spoiled if you ask me. I understand they are eager to see a D710 just to push ISO levels...even if they don't use them, but after seeing what the 5DIII costs, they better watch what they wish for....

 

[LetTheRightLensIn]

Canon does preserve the noise information more than Nikon does because of this offset, but I haven't seen any data yet that quantitates how much the noise is 'crushed' by this pseudo-noise removal in Nikon cameras.

I'd like to see what the deviation from the expected logarithmic curve one sees in the actual signals of patches of Stouffer Transmission wedge shots with these cameras. That might give some clue as to what degree shadows are crushed in Nikon RAW files, with the added benefit of a seemingly higher SNR in the dark patches due to the effective lowering of noise (which would give them higher DR estimates if you calculate DR by determining the number of stops between the brightest not-blown patch & the darkest patch that still yields SNR=1). I'm attempting to do this now with transmission wedge shots from the 5DIII & the D7000. As of now, the D7000 yields SNR>1 even at patch 42 (13.2EV), but I'm curious if shadow crushing due to Nikon's special processing of low signals ends up not representing the actual gradation between the darkest patches as well as it might otherwise.

I have no doubt that the Nikon/Sony sensors & signal processing yield better DR than Canon (Fred Miranda's comparison is rather convincing!)... I'm just curious if it's as great a difference in the real world as what the DXO testing methodology indicates.

I think it is.

OTOH I believe it is said to not be good for astro photography since it messes up proper stacking.

 

[Tcapp]

Sorry if this is a noob question but most of what I read talk of photosite size positively correlating with the DR. Somehow the cameras with better DR are not the ones that have the biggest photosites though. I understand SNR is a limiter however not sure if that is the reason for, let us say, mk3 with bigger photosites having worse DR than D800.

I took a look at the thread but could not get it clear for myself. Could anyone shed light on this? :-[

Cheers!

Photosite size will only help DR if all else is exactly the same. D800 sensor has different technology going into it than other sensors, so it overcomes the smaller photosite thing. If you took the D800 sensor, and just reduced the resolution but changed NOTHING ELSE, it might have even better dynamic range and low light performance. But you cant compare different sensors from different generations or manufacturers based on photosite size. Advancements in technology over shadow the size difference.

Its like saying that a bigger car engine gives more power. Its a true statement for the most part. But if you compare engines made today to ones produced 20 years ago, todays better technology can get more power and efficiency out of a smaller engine.

 

[sarangiman]

OTOH I believe it is said to not be good for astro photography since it messes up proper stacking.

Why would shadow crushing mess up proper stacking? It can change the noise profile, but random noise will still be removed by stacking...

I'm wondering if shadow crushing was the reason astrophotographers generally don't go w/ Nikon (aside from the fact that Canon offers cameras w/ better filters for astrophotography)? But if the read noise is extremely low, that may be irrelevant when compared to the noisy Canon images (especially when you consider FPN).

But I don't really know. Would love for an actual astrophotographer to chime in...

 

[LetTheRightLensIn]

OTOH I believe it is said to not be good for astro photography since it messes up proper stacking.

Why would shadow crushing mess up proper stacking? It can change the noise profile, but random noise will still be removed by stacking...

I'm wondering if shadow crushing was the reason astrophotographers generally don't go w/ Nikon (aside from the fact that Canon offers cameras w/ better filters for astrophotography)? But if the read noise is extremely low, that may be irrelevant when compared to the noisy Canon images (especially when you consider FPN).

But I don't really know. Would love for an actual astrophotographer to chime in...

Wouldn't it mess things up if all of the less than zero noise was gone and you wanted to average frame right near black?

Anyway I hear that they also automatically apply NR to RAW files for long exposures. Apparently people using them for astro have to play games by switching modes or shutting it off an somehow restarting the same exposure again, perhaps a few different times, to build up a long exposure without getting NR baked into the RAW.

 

[sarangiman]

Wouldn't it mess things up if all of the less than zero noise was gone and you wanted to average frame right near black?

I don't see why it would. Temporally variant noise is temporally variant whether or not its negative variation around an arbitrarily set 'black point' is clipped. You'd still get rid of it by image averaging.