Not only is it helpful for saving shots where you've messed up and under/over exposed, but it's useful for getting all those stops of DR. See the 5DII when you start pushing fill-light past 7 stops of DR -- you get banding.
It's useful (arguably extremely so) for landscape work or PJ where the extra DR is relevant.
Sp12 wrote:
Not only is it helpful for saving shots where you've messed up and under/over exposed, but it's useful for getting all those stops of DR. See the 5DII when you start pushing fill-light past 7 stops of DR -- you get banding.
It's useful (arguably extremely so) for landscape work or PJ where the extra DR is relevant.
really 7 stops wow must have a real dark images all the time...or the lens cap on.
all the cameras seem to have their best DR at 100-200 iso right so you would use
this for landscape at the correct exposure in the first place you do have time to
to this unless you need to machinegun a sunset/sunrise...sorry but still not buying
what DXO is selling here...it seems its just a way to have a pissing match between cameras with numbers as opposed to REAL pictures...
unlike the atuomotive world you rely on numbers to get the fastest time,speed
ect..
I mean 7 stops of overall DR in the scene. When you start opening up past that on the 5DII you get banding, whereas at ISO 100 the D7000 does 13 or so regularly.
Sp12 wrote:
I mean 7 stops of overall DR in the scene. When you start opening up past that on the 5DII you get banding, whereas at ISO 100 the D7000 does 13 or so regularly.
need to look at the chart again the 5D mark II has 12 stops of DR at 100 ISO
why would one intentionally take a under-exposed pic just to justify DXO results
or why would you use anything other than 100-400 ISO during the day?
Sp12 wrote:
I mean 7 stops of overall DR in the scene. When you start opening up past that on the 5DII you get banding, whereas at ISO 100 the D7000 does 13 or so regularly.
I suspect there is a completely different workflow to blame for that nonsense.
Where's the example(s) with documented workflows?
Your putative 5D II example may have been a Raw squeezed out through an 8 bit TIFF and pulverised in Photoshop, while the putative D7000 example may have been some in camera JPEG with unusual (impractical) combination of settings, such as the least amount of contrast possible.
I don't know, which is why I ask for actual examples with documented workflows.
Much depends too on how different websites mangle images that they accept for upload. I was just debugging on example on a forum where a nice clean image showed up with banding in the sky. By simply adjusting the size of the uploaded image to keep the mangler down to a dull roar, the banding was eliminated.
This is based on my personal usage of both bodies, I'll see if I can dig up any raws for you to mess with. The 5DII 'has' 12 stops of DR, but unfortunately when you start pushing the shadows it has banding noise. It doesn't show up on the 1DsIII.
To those who say that DxO is purely technical fluff and has no real world relevance, I would like to know a single landscape photographer who hasn't seen (or noticed) the limitations of the camera's DR. In my opinion, lots of the great landscape photos I see have scenes which contain a large dynamic range - much larger than the camera sensor can capture in one shot (i.e, sunsets, forests with sun rays bursting through the trees, etc). We've all been there before: You expose for the sky, but the landscape is under-exposed. You expose for the landscape, but then the sky is over-exposed. I think most people understand that the human eye has a larger dynamic range than any of the high-end DSLR's on the market today. Otherwise, if you can sell me a DSLR that can truly capture what my eye sees, then I'd love to buy one!
Landscape photographers have overcome this situation buy blending multiple exposures in post-processing, or simply using a grad filter. But sometimes multiple exposures aren't good enough (i.e, when the scene is changing quickly) or a GND isn't the best choice (i.e, when you've got a tall object that covers the vertical space in the picture). Also, even when you can use a GND filter or use exposure bracketing, these techniques can sometimes be a hassle because they take extra time, patience, planning (and money) to do so properly.
This, in a nutshell, is why DR is so important to me, as a landscape photographer who enjoys shooting high dynamic range scenes. (Maybe I should have chosen a simpler hobby?).
wickerprints wrote:
I'm not able to carry out empirical observations, so I think it will be interesting to see the relevant comparisons. We've got a lot of people saying how great Nikon's sensor performance is, and how they can push many stops in post. The question I have is whether the loss of data precision as a result of quantization outweighs any advantage conferred by more highlight latitude, and whether there is sufficient bit depth. I don't know the answer to those questions.
On paper you're right, the quantization error from base ISO pushing is significant relative to the same image taken with a higher nominal ISO. In practice however, the std deviation of noise in the lower stops (both read-noise and shot-noise) scales nicely with the fewer ADC bits available for quantization, such that the difference isn't perceptible in most cases. You can see this in practice if you look up some the D7000 ISO-less threads on dpreview.
wickerprints wrote:
Seriously, just let it go. I disagree with you, and you disagree with me.
And for what it's worth, you did not explain the significance of the oil drop experiment. Anyone who bothered to follow this inane discussion--though I doubt there are any who have--can see it in the post history. The key point you missed is that the subsequent experiments were biased toward the incorrect value provided by Millikan, and that the values trended toward the correct value over time. This could be explained only by the fact that they didn't want to contradict his work, and thus sought reasons why their values didn't match, accepting error when it favored agreement with Millikan, and rejecting it when it disagreed--i.e., confirmation bias. You mentioned NONE of this, meaning you did not understand my point. ...Show more →
hmmmm........:
"Maybe because later experimenters, for a while, tried to match his data and tried to explain away if they got a different (often more accurate) value until their data matched his? Maybe because he may have tossed away some of his outliers without a detailed enough reasoning?"?
anyway we will have to disagree on the other stuff (and not that I want to put words into someone else's mouth but I don't recall one of the developers of heterotic string theory saying anything particularly negative, for the most part, about DxO's plots on the forums, and yes, granted, not that that necessarily means anything either)
digitalbug30d wrote:
DXO is only for non-photographers obviously...I bet DXO doesnt even post pics of brickwalls,ducks,beer bottles ect...and the tried but untrue batteries in a row test gotta love that one.
Sorry I dont really care how many here with fancy degrees love that charts stuff
it really doesnt mean anything to most real photographers and I mean the ones who can take a picture and not tear it apart bit-by-bit (no pun intended)
digitalbug30d wrote:
if you take the picture at the correct exposure in the first place why the need to push it 2-4 stops? I can see 1 at the most up/down
it has nothing to do with getting the exposure wrong
or all this correct vs. incorrect exposure talk
(although if you did make a mistake, who wouldn't want to be able to rescue the image better, but this is an entirely separate matter)
maybe the scene has 16 stops of DR that you want to capture and your camera delivers 8 usable stops
digitalbug30d wrote:
need to look at the chart again the 5D mark II has 12 stops of DR at 100 ISO
why would one intentionally take a under-exposed pic just to justify DXO results
or why would you use anything other than 100-400 ISO during the day?
12 stops of engineering DR at 8MP equivalent at a certain print size etc. etc. whatever the exact criteria for their absolute numbers are and while there are plenty of scenes that won't exceed this and plenty that won't come close to exceeding this there are also plenty that will greatly exceed this (vast numbers if include all sorts of stuff that people have been trained are bad pictures only because cameras can't yet handle them)
It's pretty ignorant to believe that they are the be all and end all. As I have already said, technically it is interesting, but why on earth are you wanting to compare a 20D with a 5D3(?!) in the first place? What real world application does that have? Technically as a bit of interest to look at yes, but for real world users looking to purchase a new camera? About as much use as a chocolate teapot
