DXO uh-oh?
- Thread starter arcanej
- Start date
zigzagzoe said:I guess many of the responses were to be expected
Just out of curiosity, did you read my reply to you? I did not see a response from you about it...just curious.
zigzagzoe said:
zigzagzoe said:
Well, I'm not here to call you a troll or make any kind of assumption that I know your reasons for switching, and whether you were pre- or post-biased or anything like that. You do, to me, seem as though you made a well educated decision, and if it improved the quality of your work, which it seems is also your living, then I'm happy for you.
I do have to say that I disagree with this one point, though. I am not going to deny the DR advantage cameras built on Sony Exmor sensors have. The day I saw Fred Miranda's comparison review of the D800 and 5D III some years ago now, there was simply no denying it. I actually ran Fred's images from that review through all my denoise tools, and while I was able to recover some DR, it was at the cost of some detail particularly in the signs, and was never quite as much as the D800 had natively.
The D800 has better IQ at low ISO, plain and simple. I don't think anyone really denies that, it's kind of hard to really ignore, even if you have a brand affinity. The reason for that is Sony Exmor sensors don't throw away useful information by injecting noise into the lower echelon's of the image signal. Canon's fundamental problem is that they DO inject a lot of noise into the lower echelons of the image signal. That costs Canon IQ that their sensors (according to Roger Clark) are capable of resolving in the first place (around 15.1 stops in the case of the 5D III, which has an FWC of 68900e-, minimum read noise of 2.05e-, which leaves us with (20 * log(68900/2.05)) / 6, or 15.088 stops of sensor dynamic range. Reference: http://clarkvision.com/articles/evaluation-canon-5diii/index.html. Note, Roger Clark actually made an error in his calculation of dynamic range, he claimed it was 14.7, however he mistakenly used 2.5e- as the minimum read noise, when in actuality it was listed as 2.05e-. Hence my different result of 15.1 stops.)
The DR improvement aside, which is ultimately just opening up more bit space for usable, recoverable image data to go into, IN PRACTICE, I rarely see any major differences between landscape photos taken with the 5D III and the D800. There are a few cases where I've seen photos from the D800 that you could tell took full advantage of it's edge, but for the most part, similar kinds of landscape photos taken with different but similar cameras all ultimately have very little discernible differences in quality. Here is a random sampling off of the first 5 pages of a 500px search for "flower mountain":
http://500px.com/photo/56726642/eastern-fjords-by-boris-michali%C4%8Dek
http://500px.com/photo/39464968/spring-flowers-by-lazy-vlad
http://500px.com/photo/66442921/superstitious-twilight-by-peter-coskun
http://500px.com/photo/77534433/loowit-dreams-by-michael-bollino
http://500px.com/photo/75522371/stormy-friday-by-zsolt-kiss
http://500px.com/photo/47847750/louis%27-heaven-by-lijah-hanley
http://500px.com/photo/74066923/if-2-by-zsolt-kiss
http://500px.com/photo/75963645/miss-independent-by-rami-jabaji
http://500px.com/photo/52348676/cruel-summer-by-ryan-dyar
This sampling includes photos from the D700, D800/E, 6D, 5D III, 5D II, and even a 5D! They are all of fields of flowers in front of a mountain or some landscape. Aside from a couple images that looked overly compressed, I am hard pressed to know, just by looking at the image, which camera took which photo. Even if you start picking each image apart, if you did not have EXIF metadata telling you which camera created which photo...you could simply never tell, except in maybe the 5D case (that camera is REALLY dated now), and one photo that nicely captured the setting sun and clear detail in the foreground that I don't think would have been possible with any current Canon camera (but I still had to look at the image for a moment and think: "Wow, the sun isn't blown, and the foreground has colorful detail!").
That is generally the case, when I spend time on 500px looking at photography, while there are some rare cases of say a close up headshot portrait where you can really tell it was well lit and that the image had a ton of resolution, for the most part...I can't see DXO's scores being indicative of ANYTHING when it comes down to real-world photography. While I don't agree with Neuro's interpretation of your original post, and I think he misread it and took it the wrong way and ultimately responded poorly, I think that was kind of the kernel of his retort: DXO scores...don't really seem to be indicative of real-world photography. They may be somewhat indicative of post-process editing latitude, and in many cases having improved latitude is very possibly the most important factor for some percentage of photographers (I fully understand the "time spent on each photo" argument, especially coming from pros), but I don't think that image quality boils down to two additional stops and more resolution. I believe those are factors of IQ, two of a greedy handful, but in the end...at least, based on the photography I've personally seen when browsing around sites like 500px, 1x, Flickr, etc...I think the tool simply compliments the photographer's skill in the end. (And a better tool in the hands of a skilled photographer can, and probably will, produce better results, so the D800 in the hands of a skilled photographer can be put to more effective use than a 5D III in the hands of the same photographer.)
Just for kicks, here is a landscape photo taken with a 40D. In a blind test, I highly doubt anyone would have figured that the 40D was used to make this photo:
http://500px.com/photo/69401533/casteil-by-julien-delaval
There is actually quite a lot of DR here...from the bright sun in the upper right corner to the deep shadow detail in the foreground grass.
In my search, I did come across ONE photo taken with a D800 that just made me go WOW, and I could tell tell it took full advantage of the increased DR:
http://500px.com/photo/46785278/mt-bromo-under-the-stars-by-elia-locardi
This photo is...just...wow. This kind of photo is what makes me want to get a D800, however I'm still, at least at the moment, holding out hope that Canon will release a competitor so I can stick with my preferred ergonomics and lenses (although I would like either a much improved 16-35 or a Nikon-quality 14-24).
Upvote
0
Don Haines said:
Well, keep in mind, I usually use DSS and PixInsight for astrophotography. There is nothing DSS can do to improve your regular still photography, as it is explicitly designed to calibrate, register, and stack individual sub frames from an astro imaging sequence into a final "integrated" result. The primary cause of dynamic range improvement there is the calibration (the use of dark frames, bias frames, and flat frames to remove read noise and bias signal from the deep deep shadows of each individual light frame) and the stacking (which averages together dozens, even hundreds, of individual light frames to GREATLY reduce noise and actually increase the output resolution and detail.)
PixInsight can be used for normal photography, and some of it's tools might actually be good to run, lightly, on normal photography images. One of them would be TGVDenoise, which is absolutely phenomenal at eliminating (and I really do mean eliminating, not just reducing) high frequency noise. TGV has to be run carefully on astro images, as lower frequencies of noise are still usually present in astro images even after integration...but on a normal "terrestrial" photo, if used at the right setting, it could completely wipe out all high frequency (i.e. per-pixel) noise pretty handily. I haven't done much experimenting with that...I just poked around with a few bird photos that were taken at ISO 12800 to see what it could do, but I didn't put a lot of time into it. The vast majority of the time, I use PixInsight to reduce the noise of my astrophotography integrations, recover and enhance detail, then "stretch" them to do a kind of shadow lifting that would indeed put even the D810 to shame.
As an example, I used PixInsight to turn this:
Into THIS:
Upvote
0
That is stunning!!! I never would have guessed that beautiful image was hiding in that dark mass. I suppose it also demonstrates that while a capable camera is a good starting point, good software helps too.jrista said:Don Haines said:
Well, keep in mind, I usually use DSS and PixInsight for astrophotography. There is nothing DSS can do to improve your regular still photography, as it is explicitly designed to calibrate, register, and stack individual sub frames from an astro imaging sequence into a final "integrated" result. The primary cause of dynamic range improvement there is the calibration (the use of dark frames, bias frames, and flat frames to remove read noise and bias signal from the deep deep shadows of each individual light frame) and the stacking (which averages together dozens, even hundreds, of individual light frames to GREATLY reduce noise and actually increase the output resolution and detail.)
PixInsight can be used for normal photography, and some of it's tools might actually be good to run, lightly, on normal photography images. One of them would be TGVDenoise, which is absolutely phenomenal at eliminating (and I really do mean eliminating, not just reducing) high frequency noise. TGV has to be run carefully on astro images, as lower frequencies of noise are still usually present in astro images even after integration...but on a normal "terrestrial" photo, if used at the right setting, it could completely wipe out all high frequency (i.e. per-pixel) noise pretty handily. I haven't done much experimenting with that...I just poked around with a few bird photos that were taken at ISO 12800 to see what it could do, but I didn't put a lot of time into it. The vast majority of the time, I use PixInsight to reduce the noise of my astrophotography integrations, recover and enhance detail, then "stretch" them to do a kind of shadow lifting that would indeed put even the D810 to shame.
As an example, I used PixInsight to turn this:
Into THIS:
Upvote
0
zigzagzoe said:
You are doing what so many people do here, taking things out of context. Is a half stop DR advantage worth anything? Probably not, but if the camera with the high iso DR advantage also has a handholdable 600, which you need and are using, and it has better AF, which you also need and are using, only a fool would buy the camera system with the half stop worse DR, worse AF, and heavier lens.
You need to keep context, for you the Sony sensor offers real advantages, for many it simply doesn't, for many people other system features are far more important. But nobody here that I have ever seen has questioned the DR advantage the Sony sensor has, just how useful it is in real world shooting for them when compared to other system advantages or disadvantages.
I have some respect for what you are saying, though think you went about it wrong, but that is up to you. As for your being a pro that clears $6,000 - $12,000 per week, I find that a bit of a stretch, I also wonder why you aren't shooting what pretty much every other $500,000 a year photographer is, medium format digital, it seems to me you'd get a far higher IQ increase going to Phase One or Hasselblad with their true 16 bit files, more MP, more DR, more everything, than pissing about with any 135 toys. Oh, and didn't you realise any trade in value for your Canon kit? Last time I knew a photographer who did a large system change from Canon to Nikon it cost him less than four thousand to get a very similar setup.
Upvote
0
dilbert said:
He was talking about at high ISO, not low ISO. At high ISO, from 800 onward, the difference in stops is at best half a stop, if even a quarter stop, in favor of any given brand. The advantage with an Exmor is only about two stops or a bit more specifically at ISO 100 (it's about a stop at ISO 100, less than a stop at ISO 400).
Upvote
0
dilbert said:
So what do you do, take the f!!!!!g comment out of context, that has to be the dumbest thing I have ever seen here!.
If it takes you an hour to make an HDR you need to go on an Adobe course, even with a slow computer it takes 3-4 minutes to make a genuine 32 bit file with more than twice the DR of the Sony sensor in PS.
Upvote
0
L
Loswr
Guest
zigzagzoe said:
I've complimented DxO repeatedly for their Measurements, and I use their software for RAW conversions. I certainly do have a different attitude toward their Scores, which I find biased and misleading for the reasons I've already stated.
My attitude toward DR is that low ISO DR is not the be-all-end-all aspect of IQ for most people. There are some individuals for whom it is, and they should choose cameras other than Canon. I do take issue when those individuals use that one aspect of IQ to bash Canon as a whole. That is as silly as it would be for me going to NikonRumors Forums and posting repeatedly that Nikon isn't a good system because they lack a handholdable 600/4.
zigzagzoe said:
Go shoot. You're skilled at it, and it's more fun than debating. I am curious, though...how many of your clients notice a difference in your output between your sessions before vs. after you switched gear?
zigzagzoe said:Ok Neuro (if that is your real name), let’s begin again.
My name is Dean, as zigzagzoe isn't actually a name, which I'm sure is a huge surprise to you
My shoot is over and my last post was a little short, both in content and manners, so take it with a pinch of salt.
I'm John, it's early morning in Boston and I've got a busy day ahead, so this will be short and to be taken with a pinch of salt. I'm a research scientist (as stated in my TDP profile link in my signature). I have extensive experience in optics, microscopy (including designing and building multiphoton microscopes) and in digital image analysis.
zigzagzoe said:I don’t shoot in the dark, I don’t see the point, others can do that if they want, but 1600 is my practical limit. If you need 12800, then either get some lights, a faster lens, or better clients
Why do you assume needing ISO 12800 means it's dark? Your clients are slow, right? Flying birds are fast, even on cloudy winter days or at dawn/dusk, artificial light isn't optimal (and not feasible in many cases), I'm not aware of any 600mm lenses faster than f/4 (and certainly not ones that can be carried on a long hike through a marsh), and birds are great clients in that they never criticize your work, although admittedly they don't pay very well (at least here, maybe there's a burgeoning market for bird droppings in Oz
).ISO 12800 is also needed with my fast-moving kids in poorly lit gymnasiums. Flash isn't permitted, I'm often using an f/2 lens, and they're great clients because even though they do criticize the work, they pay in smiles, hugs and joy which are worth far more than money.
Upvote
0
zigzagzoe said:
Hey, I think the discussion has become a lot more reasonable and that's good. But I have to point out (as a fairly disinterested party - i.e. decidedly NOT a fanboy) what I perceive to be a double standard here. From what most people are saying, the low ISO difference (in most situations) is not that big. So, it shouldn't 'kill' Canon files. On the other hand, you're saying that Nikon's lenses aren't that much worse - which is true (again, from what I've heard). So an unreasonable exaggeration anti-Canon and a reasonable assessment anti-anti-Nikon. You can see why some people might find that problematic?
[quote author=zigzagzoe]
12800 wasn't even a possibility not long ago, and in film, an impossibility, and now we have photographers with such a low skill level they require it simply to get their shot.
The skill of moving with the subject, a staple of F1 and other sports photographers for decades, has seemingly disappeared in the digital age if you want me to take you bird example seriously?
[/quote]
I can't speak on motorsports, but I can say that advancing technology has indeed allowed for shots of wildlife that could not have been taken before. Better autofocus, better high ISO, and lower weight super telephoto lenses amongst other things. We can take kit to more awkward places (because it weighs less), photograph subjects at dawn and dusk, under trees, or in very overcast conditions (because of better low light capability), etc. Not that great wildlife shots weren't taken in the past, but we are freer now. That's a good thing.
ISO 12800 is an extreme, but 6400 is fairly reasonable for, e.g. small birds in flight in the shade/on a cloudy day. I find it a useful technique to overexpose (without clipping the highlights) then pull the exposure down to reduce shadow noise - so the final shot may have the appearance of a 3200 exposure, but I need a useable 6400 to get there.
[quote author=zigzagzoe]
The BBC doesn't shoot their nature docos in HD at 12800, I'm pretty sure of that, as I have the blu rays and on my 65" TV they look mostly noise free
[/quote]
Not a good comparison. First, the resolution at HD is 2MP - i.e. less than 1/10th the 5D3's sensor, let alone the D800. Downsizing to 2MP will reduce the appearance of noise. I bet some of the footage in low light is indeed at higher ISO. I doubt it would register on screen (even a massive HD tv) except with very high settings. Second, video is not the same as stills. For clean video (I am told), you want frame exposure times of 1/50-1/100 (for smooth motion). You can have motion blur in video frames with no problem, so that's okay. For a clean still of a bird in flight, you'd want between 1/1600 and 1/2500 I would say. That's a lot less light (4 or 5 stops)!
Upvote
0
I should mention also that I'd still like better high ISO. I was photographing a jay in very dense woodland once, and 12800 wasn't enough. Some will say don't bother - but that's another way of saying accept the limitations of your kit, but that's incompatible with the attitude 'nobody ever needs these settings'. Also, I most frequently shoot at f/10 now for birds - using the 500L+2x stopped down a little for sharpness. So even in the very brightest conditions I need moderate ISO (a white bird in flight in full sun still needed ISO 400!). Again, I'd like the freedom to take more shots (although an improvement in autofocus would also be required in this case).
Sorry if I'm drifting off topic a little...
Sorry if I'm drifting off topic a little...
Upvote
0
your logic is flawed. You ask people to accept that for the photography that the D800 (for a variety of reasons) is the best choice for what you wish to do. You can not accept that for other people, a 1DX and 600F4 is the best choice for what they wish to do and criticise those trying to do so as having poor technique because they may need ISO12800 to freeze motion under the shooting conditions.zigzagzoe said:Hey John, how you going. Dean here.
I think now we have a clearer picture of you, and I hope you have one of me.
As I've said before, you're a smart guy and now we know why, and thank you for the insight.
12800 wasn't even a possibility not long ago, and in film, an impossibility, and now we have photographers with such a low skill level they require it simply to get their shot.
The skill of moving with the subject, a staple of F1 and other sports photographers for decades, has seemingly disappeared in the digital age if you want me to take you bird example seriously?
That's harsh of me in some ways, and honest in others. Clean 12800 would be great (no need to get clients eyes on the same plane in low natural light for example).
It has it's applications, but let's be honest, to assume it's needed for more people that need clean high DR 100 ISO is not only a stretch, but an insult to both of us in any debate.
320 ISO, and this bike was moving very fast, and this isn't what I do really. I'm not good at tracking fast objects, but I didn't need 12800 to freeze this bike.
http://client.deanagar.com.au/0001.jpg
Now this sucker really was moving, around 180mph, faster than most birds I'd guess, on an overcast day in Melbourne.
ISO? 100 funnily enough
http://client.deanagar.com.au/sauber.jpg
This is a sad shot for me, given the state Michael is currently in, but again 100ISO, D60, 2004, his last championship winning year.
http://client.deanagar.com.au/shuey2.jpg
12800? Really? Isn't that just a technology to replace actual skill for most applications beyond scientific?
How many times does a skilled photographer need 12800? More than skilled photographers need 100 ISO?
I highly doubt it.
Anything in a studio is shot at 100 ISO, and that counts for a lot of paid work in this, or any other world.
Most shots you'd sell for actual real money are taken from 100-1600 and even 1600 is a push if we're honest for pro shots that aren't concert, wedding or editorial, and while there are commercial applications for higher ISO's, let's not insult each other by pretending they amount to more than a small percentage of amateurs that can't afford lights, or specialists trying to spot Neptune on a dark night.
I'm sure people took some great shots of birds on film John, I've seen them in books when I was as kiddy windy.
The BBC doesn't shoot their nature docos in HD at 12800, I'm pretty sure of that, as I have the blu rays and on my 65" TV they look mostly noise free
You can not ask for understanding from others without reciprocating.
and by the way, I have said time and time again on this forum, "who cares what the DR is of a blurry picture". To my way of thinking, the big strength of the 1DX is it's autofocus system. Trying to track a bird in flight, through a 600mm lens with a 1.4X teleconverter attached, where the bird is only a quarter the width of the view, is not a trivial task and take a phenomenal amount of skill and practice PLUS a kick-ass AF system.... and unlike race cars, birds do NOT fly in straight and level lines nor do they follow a repeatable pattern. Photographing race cars is a trivial problem in comparison.
Upvote
0
Hi,
Have a nice day.
Yes... with a 12.2MP FF sensor, I had no doubt that it'll kill all Canon dead at ISO 12800, but before you can get that killing picture, you need to first get the camera to focus on the kids running a dimly lit gym.... I don't put my bet on Sony A7s to be able to do that....zigzagzoe said:
Have a nice day.
Upvote
0
Which proves one thing only - you shoot in an enviable amount of light.zigzagzoe said:
Oh - and a moto-x bike bottoming out in a berm is not "moving very fast". 30 mph tops, and a much bigger, more predictable target than a small bird in flight.
And it's soft and out of focus. Maybe a higher ISO would have been a good idea, eh?Now this sucker really was moving, around 180mph, faster than most birds I'd guess, on an overcast day in Melbourne.
ISO? 100 funnily enough
I mention these points simply to underline the idea that "professional" does not automatically imply "expert", and these images demonstrate the point very nicely.
And again - if you need 12800 ISO to get a shot, "actual skill" will not and cannot make the difference. If the light is crap, and you need a high shutter speed, a high ISO is the only option once you're "out of lens".
Based on what we're seeing in this and all other similar threads on this subject, photographers - "skilled" or not - seem to have more need, all told, for better high(er) ISO performance, because most aren't shooting obliging, static, stable, predictable lumps of geography from a tripod.
Irrelevant, and a niche example: most photography is not done at 100 ISO in a studio, is it?
Really, they did not. They took the best pictures the technology of the day allowed them to take, and for the most part at a quality that today wouldn't merit a second look.
Which is again an irrelevant deviation from the points under discussion.
It's neither "fanboy" nor "horse S*@t" - it's a demonstrable fact.
Again: nobody is denying that Nikony has a low ISO DR advantage over Canon, the dispute is over its significance (which for the most part, for most photographers, is small); and its necessity in order to achieve "The Shot", and the case for that is laughably weak - even taking into account the pitch-black bar image that Horshack/Snapsy has posted up in the past: impressive that it could be recovered from black to usable in post, but - really - just use a higher ISO.
Upvote
0
OK.... I accept that I have a low skill level. Now tell me what I should have done here. This is the unedited jpg out of the camera, and yes, I do shoot in RAW....zigzagzoe said:
ISO12800, F1.4, 1/25th of a second, in a venue where flash or other lighting is not permitted. What should I have done to have avoided using ISO12800?
Attachments
Upvote
0
L
Loswr
Guest
zigzagzoe said:
Decent example shots. Your subjects are pretty large and let you approach relatively close. I wonder if that makes a difference? I like the dirt bike shot. It must have been a really dark day for dirt biking, what with the clear blue sky and the sun on the bike and all. Your shutter speed of 1/320 s was several stops slower than needed to freeze bird wing motion, too. Guess what? I don't shoot at ISO 12800 on sunny days.
The racing shots are less impressive, at least from a framing standpoint. Or maybe you were going for a Luke Skywalker landspeeder look in the first one? It wasn't dawn or twilight at the racetrack, was it? Again a shutter speed on the slow side, since you were panning and showing wheel motion, which is usually desirable in racing shots. I sometimes use shutter speeds as slow as 1/1250 s to leave some motion in the wingtips of large birds, or as slow as 1/2000 s for small birds. Freezing wing motion means 1/2500 s to 1/30000 s (the latter being achievable with a feeder setup and several Speedlites at minimum power).
Like I said, decent example shots...even if they don't illustrate anything truly relevant to the discussion.
zigzagzoe said:The BBC doesn't shoot their nature docos in HD at 12800, I'm pretty sure of that, as I have the blu rays and on my 65" TV they look mostly noise free
That red herring should have been thrown back, it's not big enough to make a meal. Surely you understand that video is generally intended to show smooth motion, and shutter speeds used to freeze fast motion are entirely different?
Upvote
0
My Dad was a police photographer. He did not have the option of saying "let's keep the debris and bodies all over the highway until a bright sunny day comes along". Sometimes you have to shoot NOW! with what you have under the conditions you are dealt... and you have to make do with the equipment you have. Setting up studio lighting is not always an option either.....dilbert said:zigzagzoe said:...
I'm sure people took some great shots of birds on film John, I've seen them in books when I was as kiddy windy.
The BBC doesn't shoot their nature docos in HD at 12800, I'm pretty sure of that, as I have the blu rays and on my 65" TV they look mostly noise free
The main difference here is that the BBC and National Geographic photographers aren't idiots that think ISO12800 is going to make up for shooting when the conditions favour the photographer. They're also patient in that they wait for the weather to give them the light they need or plan their photography so that the odds are in their favour.
I don't pretend to understand why neuro thinks it is a good idea or useful to shoot flying birds in cloudy conditions (or worse) but it sounds to me like he's confined his shooting to a corner where nobody can produce a better camera system for what he does than the one he owns (or so he thinks.)
Why would anyone want to shoot birds flying when it is cloudy I don't know. Any colours that might be in/on the birds are going to be greatly subdued and unless you're shooting B&W, isn't the goal to get good colour?
Upvote
0
