Luminous Landscape Forum
Raw & Post Processing, Printing => Digital Image Processing => Topic started by: Frans Waterlander on April 15, 2020, 07:53:08 pm
-
I want to preface this discussion by excluding the areas of HDR imaging that some described as "clown vomit": exaggerated effects way outside of realistic and indoor real estate photography with its stark differences in brightness between the indoors and outside. Maybe there are other specialized areas needing bracketing and HDR programs that fall outside of this discussion. So my ramblings should apply to general photography.
DSLR camera dynamic ranges have improved significantly over the years as have tone mapping capabilities of RAW converters like Adobe's ACR. So, it seems to me that there is less of a need for bracketing and HDR programs for most shooting conditions.
I have run experiments using the Grand Canyon, Manhattan and Queen Mary sets of bracketed RAW images on the DVD that came with Christian Bloch's HDR Handbook 2.0.
Step 1 was to merge those images in Photomatix 6.1 and select the most pleasing (to my eye) Realistic renderings. Step 2 was to import the single, middle-exposure images into Photoshop Elements and tweak them in ACR and after RAW conversion to come as close as possible to the results obtained with merging the sets of bracketing images in Photomatix.
The results I got with Elements were equal to the results with Photomatix in terms of image quality, visible noise without pixel-peeping, colors, clarity, etc.
What say you?
-
Answer: Yes, many times we do. When we wish to fully capture scenes that exceeded a cameras limited DR.
DSLR camera dynamic ranges have improved significantly. Not enough to capture the DR of many scenes.
Shown here, explained here:
http://www.lumita.com/site_media/work/whitepapers/files/pscs3_rendering_image.pdf
-
Yes, we still need them, maybe less often than a few years ago.
Even for a camera with improved dynamic range, depending on the illuminant PSD, you may end up with a severely underexposed raw channel.
-
I agree with Andrew, even shooting in "raw" constantly.
When I'm on the road and time is limited but the dynamic range is vast, I'll still do a three-shot hand-held HDR capture and hope they will align well in post. When I have the luxury of a tripod, it's an easy shoot. Last week, it was under "Hail Mary" lighting up 40 feet of scaffolding and where I couldn't put a tripod nor use the lowest ISO. So, it was a do a 3-shot bracket, -2, N & +2 and shoot away.
Using lots of noise-reduction and ACR's HDR did the job and both my clients and I were happy with the result.
IMO, HDR is useful for not only architecture but also for landscape at the edge of the day and many times better, more natural looking than using a grad ND in many ways.
-
I'm with Andrew as well.
While digital cameras are waaay better these days than 10 years ago there are still scenes where HDR gives me more latiude to work with. I use Adobe HDR from Lightroom and Photomatix Pro 6.2 and in some rare cases Nik HDR Efex Pro 2. I like to shoot in a wide range of lighting and yes, sometimes I take -2,0,+2 exposures as an insurance. A lot may depend on what you're doing in post processing. I don't always want to make an image that is a strict representation of the scene I saw and the latitude in the extremes is a boon at times.
-
What I find silly is that some cameras still have options to bracket in 1/3 EV up to 1 EV increment, a few 2EV.
Nowadays the minimum step might well be 2EV with the option of 4EV
-
Yes here as well, and the good news is that with most software these days it's not difficult to get natural looking results.
I do agree with Francisco that 0.3ev increments are not needed but they can still be useful to ensure that the right exposure (a single exposure) is captured in a difficult situation.
-
What I find silly is that some cameras still have options to bracket in 1/3 EV up to 1 EV increment, a few 2EV.
Nowadays the minimum step might well be 2EV with the option of 4EV
+1
Victor B
-
Wether you are rendering one file or 10, if the dynamic of the scene hasa high contrast, you will need to remap correctly the values to make an hdr rendering. That means you need to compress a wide bunch of datas in a smaller one (at least your screen), and make it pleasant (enough contrasty, not to flat, not too fake, etc. There is a lot of method for tone mapping, many algorithms propose to do it automatically, but you can also do it by hand. Anyway, it's still an hdr rendering.
In my practice (mostly indoor architecture), I'm still doing bracketing on tripod. But most of the time I use only one picture : I use the maximum exposed picture to minimize noise, and get the wider dynamic possible, without clipping highlights of course.
Only sometimes I need to use a clipped picture, to use it for shadows : not often for shadow recovery, most often mainly for reducing noise in darks.
From my needs point of view, sensors have made enough progress that noise in darks became acceptable in mid-challenging situations. I save times using only one picture.
Of course, if you like taking pictures in forest clearings, maybe it won't be enough for your needs. All theese considerations relies first on scenes datas.
-
I'll lend my voice to the chorus of "Yes, we do." My Canon 5D Mk IV has a very broad dynamic range, but it is still not equal to the human eye. There are situations where there is too much dynamic range in a scene for the camera to capture adequately.
Now, I do tend to use a three-shot bracket of +2, 0 -2 in such situations. Sometimes, even a five-shot bracket.
Cheers,
Mitch
-
Yes here as well, and the good news is that with most software these days it's not difficult to get natural looking results.
I do agree with Francisco that 0.3ev increments are not needed but they can still be useful to ensure that the right exposure (a single exposure) is captured in a difficult situation.
Can't colors change as well even at 1/3 over or under exposure? So bracketing at 1/3 would assure the correct colors important for product shoots.
-
What I find silly is that some cameras still have options to bracket in 1/3 EV up to 1 EV increment, a few 2EV.
Nowadays the minimum step might well be 2EV with the option of 4EV
Well, not completely. This is indeed correct if you’re merging 2 or more images in post.
However, there are many cases you’re unable to do so, for instance due to movement differences between the shots. Here you want to get everything right in one shot and then bracketing with half a stop or one stop may become handy. ETTR so to speak.
If we would only have RAW histograms…..
Regards,
Jaap.
-
If we would only have RAW histograms…..
Regards,
Jaap.
I wonder how the camera manufacturer would accomplish that. Would it require demosaicing the scene in the camera?
-
I wonder how the camera manufacturer would accomplish that. Would it require demosaicing the scene in the camera?
Some Phase backs provide such raw Histogram as can some Canon's with:
https://www.magiclantern.fm/
However it's done under the hood, it's doable and sadly rare. Because for some strange reason, most manufacturers don't want to (not can't). 😩
-
I wonder how the camera manufacturer would accomplish that. Would it require demosaicing the scene in the camera?
It is actually the other way around. In order to show the sRGB or Adobe RGB histogram, the raw file is processed in camera. A raw histogram would not require it.
-
Really appreciate all your inputs. Would you care to comment on my experiments with Christian Bloch's bracketed shots?
-
Really appreciate all your inputs. Would you care to comment on my experiments with Christian Bloch's bracketed shots?
Well first of all, my first edition (2007) of the book came with no DVD with images and since you didn't (nor should you) attempt to upload them for some of us, no, I can't comment on something I (we) haven't seen from you nor can we without the images, try ourselves.
-
Would it require demosaicing the scene in the camera?
of course it is not required ... did you pause and think before writing this ? raw histogram is about the DNs from sensels per CFA filter ("foveons" of course different) .. granted you can instead also use histogram after demosaicking too - provided you did not apply WB and other modifications (any kind of LUTs)... but histograms are quite useless (raw or not) - blinkies or zebras are much better because they show you the clipping (or might show you how close you are to) in place in the frame - or you can have some monochromatic "view mode" of the scene where lightness can be used to show you some map of how close you are to the clipping across the whole frame with clipped (in one or more raw channel) areas blinking in some color (add a different color blinking for areas where S/N is below a certain threshold too)... probably even better than regular blinkes/zebras we have now
-
of course it is not required ... did you pause and think before writing this ? raw histogram is about the DNs from sensels per CFA filter ("foveons" of course different) .. granted you can instead also use histogram after demosaicking too - provided you did not apply WB and other modifications (any kind of LUTs)... but histograms are quite useless (raw or not) - blinkies or zebras are much better because they show you the clipping (or might show you how close you are to) in place in the frame - or you can have some monochromatic "view mode" of the scene where lightness can be used to show you some map of how close you are to the clipping across the whole frame with clipped (in one or more raw channel) areas blinking in some color (add a different color blinking for areas where S/N is below a certain threshold too)... probably even better than regular blinkes/zebras we have now
Of course I thought about. That's why I phrased it as a question. I didn't know. I bow to your wisdom.
I'm glad you got the opportunity to show off though.
-
Of course I thought about. That's why I phrased it as a question. I didn't know. I bow to your wisdom.
I'm glad you got the opportunity to show off though.
+1
-
The answer to that depends on the camera used, and the scenes shot.
If you have an IQ4 with Frame Averaging (https://www.dtcommercialphoto.com/the-phase-one-iq4-now-has-automatic-frame-averaging/) and Dual Exposure+ (https://www.dtcommercialphoto.com/dual-exposure/) then the answer is basically: no, you don't.
Except for some *really* contrived situations the dynamic range of that system easily exceeds the contrast of even traditionally challenging scenes (e.g. dark subject matter backlit against bright subject matter).
If you're using a 10 year old Canon Rebel then the answer is basically: yes, you do. Even moderately challenging scenes will easily outpace the range that camera can capture in a single frame.
[assumptions are that you want a file that contains usable detail throughout the subject, leaving it up to you how to render the tonality of the scene, without fear of introducing ugly noise or areas of otherwise poor quality such as banding or lower color fidelity]
-
Can't colors change as well even at 1/3 over or under exposure? So bracketing at 1/3 would assure the correct colors important for product shoots.
Colours don't change with exposure, unless the original RAW files are processed with poor (very poor indeed) software. In fact the most accurate colours are achieved through ETTR, no matter if the final exposure will be lower. The only difference between differently exposed shots is SNR.
This is a seamless composite made from two RAW files 2EV apart, the joints are undetectable in the whole image because capture exposure doesn't alter colour nor contrast:
(http://guillermoluijk.com/article/ettrcolor/indetectable.gif)
And these are the colours on captures taken 4EV apart:
(http://guillermoluijk.com/article/ettrcolor/0ev.jpg)
(http://guillermoluijk.com/article/ettrcolor/4ev.jpg)
http://translate.google.com/translate?u=http%3A%2F%2Fwww.guillermoluijk.com%2Farticle%2Fettrcolor%2Findex.htm&langpair=es%7Cen&hl=EN&ie=UTF-8
Regards
-
I wonder how the camera manufacturer would accomplish that. Would it require demosaicing the scene in the camera?
A histogram is just a count of how many pixels in the RAW file are given each possible numeric value after the A/D conversion. So you can calculate and display a RAW histogram very easily without demosaicing anything. This is IMO the most beautiful, compact and useful to the photographer RAW histogram:
(https://assets.change.org/photos/5/ic/ze/EViCZEQriNMtQfr-800x450-noPad.jpg)
The reason I find it the most useful to the photographer is that it is plotted in log scale, so the width of every vertical division corresponds to exactly 1 stop in exposure. This has the following advantages:
- The histogram doesn't change in shape when modifying exposure, it justs shifts left or right:
(http://www.guillermoluijk.com/misc/histogram.gif)
- It allows to know exactly how much you need to overexpose to achieve ETTR without trial&error (in my example the RAW file is perfectly ETTR'ed, it wouldn't even admit an extra +1/3EV)
- It allows to calculate scene's dynamic range at a glance (even if scene's DR is higher than the sensor's DR!). This scene spreaded over 8 stops of DR.
Regards
-
Responding to the original question, I think there are still situations where more than one shot are still recommended or even we could say necessary. However in my opinion the limit, i.e. the point at which more than one shot won't be recommended/necessary anymore thanks to sensor's DR is nearing. There will always be real world scenes with more DR than any sensor because real world contrasts can be so huge, but we'll probably won't want to do more than one shot on those scenes because tone mapping such a high contrast in a good looking final image will be very dificult, if not impossible.
- A scene with 12 stops of DR, no matter if it comes from a single shot or more, can be quite easily tone mapped on a monitor or print in a pleasant way.
- A scene with 14 stops of DR starts to be difficult to render on any output device in a pleasant way.
- A scene with 16 stops of DR is nearly impossible to render properly on any output device, the compression will lead to an unreallistic look or loss of local contrast. If our camera has 14 stops of photographic DR, we'd rather sacrifice a couple of the scene's 16 and just render 14 stops, than doing an extra shot +2EV apart with the idea to merge them. So it won't make any difference capturing those extra 2 stops (either with a couple of shots with the 14 stops camera or using a hypothetical 16 stops camera).
Eventually at some point increasing DR on sensors will become useless, and the reason is not we can't find real world scenes with such a high contrast but because it will be almost impossible to map such a large DR on any output device in a pleasant way. The compression required will make those images look dull or unreallistic.
Regards
-
Responding to the original question, I think there are still situations where more than one shot are still recommended or even we could say necessary. However in my opinion the limit, i.e. the point at which more than one shot won't be recommended/necessary anymore thanks to sensor's DR is nearing. There will always be real world scenes with more DR than any sensor because real world contrasts can be so huge, but we'll probably won't want to do more than one shot on those scenes because tone mapping such a high contrast in a good looking final image will be very dificult, if not impossible.
- A scene with 12 stops of DR, no matter if it comes from a single shot or more, can be quite easily tone mapped on a monitor or print in a pleasant way.
- A scene with 14 stops of DR starts to be difficult to render on any output device in a pleasant way.
- A scene with 16 stops of DR is nearly impossible to render properly on any output device, the compression will lead to an unreallistic look or loss of local contrast. If our camera has 14 stops of photographic DR, we'd rather sacrifice a couple of the scene's 16 and just render 14 stops, than doing an extra shot +2EV apart with the idea to merge them. So it won't make any difference capturing those extra 2 stops (either with a couple of shots with the 14 stops camera or using a hypothetical 16 stops camera).
Eventually at some point increasing DR on sensors will become useless, and the reason is not we can't find real world scenes with such a high contrast but because it will be almost impossible to map such a large DR on any output device in a pleasant way. The compression required will make those images look dull or unreallistic.
Regards
+1
-
Colours don't change with exposure, unless the original RAW files are processed with poor (very poor indeed) software.
If noise increases, color sensitivity will decrease. Colors can shift between different exposures, as you wrote in your article, and as you show here (http://www.guillermoluijk.com/article/ettrcolor/noisebluetint.jpg).
If it's not so much visible with a big photosites/sensors, it can be quite obvious with small photosites/sensors.
Moreover, a not well controlled denoising or a bad black level substraction can affect linear datas and can be rather harmful for colors, especailly if underexposure is too strong, wich can often happen in high dynamic scenes.
Consequently, multi exposure bracketing will often improve not only exposure, noise and details, but also colors separation, especially in darks.
-
If noise increases, color sensitivity will decrease (...)
Yes, but this is not the "colours change with exposure" statement under discusion. 4EV is not 1/3 EV:
Can't colors change as well even at 1/3 over or under exposure? So bracketing at 1/3 would assure the correct colors important for product shoots.
Regards
-
I would say if your shooting for the web and your capturing interior and you charge 5.00 per shot use an IPhone
Any great retouched would love to have 9 images 1 stop. Apart and even more depending on the scene to deliver excellent work that you would hang on a wall.
There is just many more opportunities to pull out detail in the shadows with no noise using feathered luminosity masks.
I also believe under exposing certain colors give the color more saturation.
A raw histogram by channel with a way to correct it would be ideal. For oceanscapes at sundown.
Phptomatrix has its uses for some images as a layer brought back into photoshop and blended.
Hard in Lightroom basically SUCKS!
Certain programs have some really nice feature that do heavy lifting with ease.
But in the end you need to develope your own style it’s more about the finished piece.
Any raw file SOC needs help, but the jpegs are cooked well, ha ha
Best of luck
-
I'm a big HDR fan. Here is an example. When I saw this scene I thought the dynamic range would be great. However, my old Pentax spot meter said the brightest part of the curtains was EV 15.3 and the cat was EV 7. So only 8.3 stops. I made a 5 shot bracket, 1 stop increments.
The first image is an HDR merge by SNS-HDR. The second image is my best attempt at processing only the middle exposure in ACR and Photoshop. The images are from a Canon EOS R. Different sources rate the dynamic range of that camera between 13 and 14.1. It seems to me that there is a big difference between theory and practice.
If anybody is interested, I'll Dropbox the raw files. I'd like to see if anyone can come close to the HDR image using only a single exposure.
(https://photos.smugmug.com/photos/i-HXVRSBh/0/42b76ca6/O/i-HXVRSBh.jpg)
(https://photos.smugmug.com/photos/i-c7PzckF/0/1ed7f7ed/O/i-c7PzckF.jpg)
-
The second shot looks like you have a black cat. The first looks like a gray cat. :)
Actually, I think the first would be slightly better if you darkened the shadows a little. The wood on the furniture seems a little to light for the way the room is lit. It would be more natural, to my eyes anyway. Shadows are shadows. While the 2nd is too dark, we really don't have to see too many details in them. We don't expect to in real life either.
-
The second shot looks like you have a black cat. The first looks like a gray cat. :)
They both look black to me; is your browser color managed? Anyway, the rendering is subjective and up to the image creator to decide.....
-
These images were made this morning. When I finished the HDR I loaded it onto my tablet and took it to the scene. The cat was still there. I looked back and forth, tablet to scene to tablet. Of course, as I scan the scene my eyes adjust. That's the point. Eyes have a narrow point of view. As I look at the table lower left, my eyes adjust and it looks very close to the HDR image. Ditto the cat.
If anything, the white curtains in the image are a bit darker than I see in the real scene. However, I have a blown pupil in my left eye and it can't close down very much. So bright things always appear brighter unless I close the left eye. Then, the curtains in the scene and the image are much closer.
HDR images, if well done and conservative, are always a better match to what my eyes see.
-
Very nice shot and a good demo of well executed HDR. The cat is clearly black in both but with a little more detail in the 1st. Also more detail in the front/shadow of the dresser. I much prefer the 1st.
I rely on photonstophotons for dynamic range info and he lists the Canon R as having ~10.5 at base ISO. You don't mention what ISO you took the images at but PDR drops off very quickly with increasing ISO, by 400 for example it's down to ~9.5
-
They both look black to me; is your browser color managed? Anyway, the rendering is subjective and up to the image creator to decide.....
Yes it's a NEC PA242W calibrated with Spectraview II. Sure the 1st cat is black. But the light in the hairs give a gray appearance. You can see the whiskers and differentiate the bottom of its chin from its breast. You don't see those things in the 2nd.
What's interesting is that while the black items are darker in the 2nd, the white sheet is whiter in the 2nd rather than darker than the 1st. How do you do that?
Of course, the final image is the way the creator wants to present it. It's very individual.
-
The answer to that depends on the camera used, and the scenes shot.
If you have an IQ4 with Frame Averaging (https://www.dtcommercialphoto.com/the-phase-one-iq4-now-has-automatic-frame-averaging/) and Dual Exposure+ (https://www.dtcommercialphoto.com/dual-exposure/) then the answer is basically: no, you don't.
Except for some *really* contrived situations the dynamic range of that system easily exceeds the contrast of even traditionally challenging scenes (e.g. dark subject matter backlit against bright subject matter).
If you're using a 10 year old Canon Rebel then the answer is basically: yes, you do. Even moderately challenging scenes will easily outpace the range that camera can capture in a single frame.
[assumptions are that you want a file that contains usable detail throughout the subject, leaving it up to you how to render the tonality of the scene, without fear of introducing ugly noise or areas of otherwise poor quality such as banding or lower color fidelity]
Can someone who has used the Phase One frame averaging reference above and does not have a stake in selling photographic equipment comment on the effectiveness of this technology?
I will mention that that are other ways to combine a range of different exposures than with an HDR program or some process that is internal to a camera or digital back, and this can provide more flexibility for certain purposes.
-
Yes it's a NEC PA242W calibrated with Spectraview II.
That is your display. I asked about your browser.
Look at the actual RGB values coming off a display with a color managed browser of the black cat. It's damn near as black as black can be, NOT gray:
-
That is your display. I asked about your browser.
Look at the actual RGB values coming off a display with a color managed browser of the black cat. It's damn near as black as black can be, NOT gray:
I think there is some miscommunication here. I think he is saying that the shadows are too open for his taste for the top shot and he would like to see them brought down more. Maybe something between the two shots for the shadows?
-
I think there is some miscommunication here. I think he is saying that the shadows are too open for his taste for the top shot and he would like to see them brought down more. Maybe something between the two shots for the shadows?
Bottom Cat: Black hole, no detail. Top Cat, lots of detail, still looks black to me, not gray. And the output values to my display (Which the Digital Color Meter provides, not the values of the image itself) show the top cat has a very deep black (5/5/5) while still having all kinds of detail invisible on the bottom cat; on this display with a color managed browser. I can't even see the bottom cat's nose but on the top, the nose and whiskers.
IF the desire of the image creator is to show us the black cat, the bottom image doesn't. The top does.
The wood shadow detail of the table (if important) is night and day between the two.
-
I think there is some miscommunication here. I think he is saying that the shadows are too open for his taste for the top shot and he would like to see them brought down more. Maybe something between the two shots for the shadows?
Exactly. That's what I was saying. But as they say in recipes, season to taste.
-
So, what happens if you pull up the shadows of the second (single) shot? Wouldn't it come close to the first shot? Just wondering. As I said before, when I played around with Christian Bloch's bracketed shots in Photomatix Pro I could make the best single shot look just as good as the HDR rendered bracketed shots.
-
So, what happens if you pull up the shadows of the second (single) shot? Wouldn't it come close to the first shot? Just wondering. As I said before, when I played around with Christian Bloch's bracketed shots in Photomatix Pro I could make the best single shot look just as good as the HDR rendered bracketed shots.
It depends upon the range of exposures. I can't speak to the examples in the Bloch book, since I have not seen them. As a general rule, with a very contrasty scene, even if you can process a single exposure so that it looks approximately the same as an HDR image, if you examine the shadows at high magnification, you will typically see much less noise there, assuming you have shot and processed the HDR image properly. This might not matter much if the image will be viewed small; but the larger the image is viewed, the more likely any significant shadow noise will become apparent to the viewer.
In any case, as some have said above, there are still going to be scenes that are so contrasty that even the most advanced sensors will not be able to capture them with sufficient highlight and shadow detail in a single exposure, to make a satisfactory image.
-
If anybody is interested, I'll Dropbox the raw files. I'd like to see if anyone can come close to the HDR image using only a single exposure.
I'd be interested in playing with those files if you post a link. I have a new version of Capture One to play with, so it might be an interesting exploration for me..
-
The first image is an HDR merge by SNS-HDR. The second image is my best attempt at processing only the middle exposure in ACR and Photoshop. The images are from a Canon EOS R. Different sources rate the dynamic range of that camera between 13 and 14.1. It seems to me that there is a big difference between theory and practice.
Not really, the difference is between SNS-HDR processing and your processing. As long as you capture the scene's entire dynamic range, the only difference you will see in the result using a different number of shots is because of software processing, not because of the number of camera shots itself.
Some software (SNS-HDR, Photomatix,...) will produce a different result depending on how many images you provide them with different exposure, but not on the number of captures done. In your cat's scene you could build 5 replicas at 1EV intervals of the single shot containing all the DR, and the result would have been the same if you manage to fool SNS-HDR to process them (you would probably find your cat a bit noisier, but this was expected and has nothing to do with the strong differences in local contrast, tones and specially shadow lifting you are showing here).
I made the following experiment: 7 shots at 1EV intervals: 1-2-3-4-5-6-7, and built two HDR linear merges (http://translate.google.com/translate?u=http%3A%2F%2Fwww.guillermoluijk.com%2Ftutorial%2Fzeronoise%2Findex.htm&langpair=es%7Cen&hl=EN&ie=UTF-8): one using all seven shots, and one using only 1-3-5-7 shots. From the two resulting merges I generated 4 results with Photomatix (please ignore the technical quality or scene interest) using a different number of replicas of the merge:
- A: 7 shots merge 1-2-3-4-5-6-7, 7 image replicas at 1EV intervals
- B: 4 shots merge 1-3-5-7, 7 image replicas at 1EV intervals
- C: 4 shots merge 1-3-5-7, 4 image replicas at 2EV intervals
- D: 7 shots merge 1-2-3-4-5-6-7, 4 image replicas at 2EV intervals
The result A = B, and they differ from C = D. In other words, those results in which Photomatix was provided the same number of images differently exposed (either 4 or 7), no matter if they were produced from a different number of original shots, are identical. And when a different number of images were provided, no mater if they came from the same number of RAW files, are different.
(tomas = camera shots, imágenes [copias] = images loaded in Photomatix with different exposure)
(http://www.guillermoluijk.com/misc/photomatix1.gif)
100% crop on dark area:
(http://www.guillermoluijk.com/misc/photomatix2.gif)
Conclusion: the key variable here is number of images (at different exposures) provided, not the number of shots (RAW files) done with your camera. The number of captures needed depends on the scene while the effect of the number of images provided to your tone mapping software depends on how your software tone mapping algorithm works. It's nothing to blame at sensor behaviour or number of shots.
I would add, because I know this is possible, that a good tone mapping software should provide the same result no matter of the number of images it is provided, as long as the whole scene's DR is captured. Unfortunately this is not the case for Photomatix, and probably SNS-HDR (I didn't try it).
Disadvantages of more shots:
- Loss of detail because of micro missalignments of shots in overlapping areas (don't consider software alignment will work perfect because missalignments will never happen at 1 pixel multiples)
- Prone to ghosting artifacts with moving subjects (water, trees, clouds passing by the Sun casting moving shadows,...)
- Waste of extra memory resources (card, HD)
Final practical suggestion: those who use automated tone mapping software are probably better served with just one or a couple of shots (3-4 EV apart), and feed their software with evenly spaced replicas of them, rather than using the camera as a machine gun. Finding out the optimum number and exposure spacing of images is a matter of trying.
Regards
-
Here you go, black cat pics. Have fun...
https://www.dropbox.com/s/hxycgscpesj4pt6/Cara42320-010v1.dng?dl=0
https://www.dropbox.com/s/3pci8y06w51cml3/Cara42320-009v1.dng?dl=0
https://www.dropbox.com/s/zy49zs6z8k1ig2s/Cara42320-008v1.dng?dl=0
https://www.dropbox.com/s/i608fxvt5402i99/Cara42320-007v1.dng?dl=0
https://www.dropbox.com/s/bt02s3hzcvpi2er/Cara42320-006v1.dng?dl=0
-
The closest to perfect ETTR was Cara42320-008v1.dng (Cara42320-007v1.dng has lower exposure so it's unnecesary because the former already preserved the highlights).
The RAW data in Cara42320-008v1.dng allows to very accurately calculate the scene's dynamic range:
(http://guillermoluijk.com/misc/blackcatrawhist.png)
As you can see your scene had RAW data spread over 11 stops of DR on each channel. Bearing in mind that for some reason you shot at ISO400 instead of ISO100, and your camera has a photographic DR at ISO400 (https://www.dxomark.com/Cameras/Compare/Side-by-side/Canon-EOS-5D-Mark-III-versus-Canon-EOS-R___795_1262) around 10,8 stops, you were past the limit and probably the noise in the cat (which corresponds to the darkest area of the scene) would be unnaceptable in a large copy. In other words, if you wanted a clean cat your camera cannot properly handle this scene using a single shot, and at least two are recommended.
Your 8,3 stops estimation of DR using the spot meter was totally wrong. I have to say I am not surprised because spot metering over relatively small subjects leads to frequent errors. A genuine 8,3 stops scene would be no real challenge for your camera in a single shot, even at ISO400.
(http://guillermoluijk.com/misc/blackcat.jpg)
(if you are surprised of so much noise is because most RAW developers apply noise reduction and we are used to it. This is what the cat really looks like with 0 NR)
BTW I guess you did some cloning on the curtains in the HDR version to hide the visible wooden part of the window, since all of the RAW files show this wooden piece.
Regards
-
So, how do I open those dng's in Photoshop Elements?
-
So, how do I open those dng's in Photoshop Elements?
With ACR of course.
-
With ACR of course.
I know that. But what would be the steps to do that?
-
I know that. But what would be the steps to do that?
File.... Open....
-
Guillermo,
Thanks for the analysis. Where is that histogram from? Does not match anything I get from Raw Digger.
Nice to know the spot meter can be so wrong.
Yes, my HDR version was a keeper so I did some cloning to clean up, including the gap in the curtains.
-
Guillermo,
Thanks for the analysis. Where is that histogram from? Does not match anything I get from Raw Digger.
Nice to know the spot meter can be so wrong.
Yes, my HDR version was a keeper so I did some cloning to clean up, including the gap in the curtains.
I calculated it on my own from your RAW data (extracted using DCRAW and plotted with Histogrammar (http://guillermoluijk.com/tutorial/histogrammar/index.htm)). The pure RAW histogram, which should be similar to what you get from RawDigger (in log scale), looks like this:
(http://guillermoluijk.com/misc/blackcatrawhist.gif)
(http://guillermoluijk.com/misc/blackcatrawhist.png)
Being totally correct, it's difficult to interpret because those spikes in the shadows due to the quantization make it difficult to find out how much information there is on every stop.
But there is a statistical trick to have a much more obvious (softer) histogram: averaging RAW data fills all levels in the histogram with the amount of light they received (or a very precise approximation of it thanks to statistics). The histogram I showed previously was a 8x8 pixels averaging. This allows to calculate scene's DR with cameras having less DR that the contrast of the scene. Actually the limit is only the RAW bitdepth.
Regards
-
File.... Open....
In case you didn't realize, I've never tried to open a file from someone else's dropbox, so I could use some hints other then File... Open...
-
In case you didn't realize, I've never tried to open a file from someone else's dropbox, so I could use some hints other then File... Open...
How would anyone know your lack of experience in opening a DNG into ACR until you tell us your lack of experience?
Do you not know how to download the DNG in the first place or you figured that out, have the DNG, but don't know how to open it in ACR/Elements???
Do examine your specific question (copy and paste):
So, how do I open those dng's in Photoshop Elements?
WHY do you make it so difficult to help you? ;)
-
You really don't need much more than that Frans. Opening a DNG will open ACR which you mentioned you know (and are familiar with?).
If you want to do an HDR with some/all images you can select them together and choose open, they will open together in ACR and you can choose to blend them with HDR after selecting them all, or concentrate on one of them.
-
You really don't need much more than that Frans. Opening a DNG will open ACR which you mentioned you know (and are familiar with?).
If you want to do an HDR with some/all images you can select them together and choose open, they will open together in ACR and you can choose to blend them with HDR after selecting them all, or concentrate on one of them.
I just don't know how to get those dng files out of the dropbox and onto my computer as RAW files and open them in PSE. I can click on them and then save them, but then they are jpegs, not RAWs.
-
I just don't know how to get those dng files out of the dropbox and onto my computer so I can open them in PSE.
I don't have to guess that's right.
”Dog helps those who help themselves” 😌
https://help.dropbox.com/
-
I just don't know how to get those dng files out of the dropbox and onto my computer as RAW files and open them in PSE. I can click on them and then save them, but then they are jpegs, not RAWs.
Use the download button at the upper right of the dropbox page, as shown in the attached image
-
I calculated it on my own from your RAW data (extracted using DCRAW and plotted with Histogrammar (http://guillermoluijk.com/tutorial/histogrammar/index.htm)). The pure RAW histogram, which should be similar to what you get from RawDigger (in log scale), looks like this:
(http://guillermoluijk.com/misc/blackcatrawhist.gif)
(http://guillermoluijk.com/misc/blackcatrawhist.png)
Being totally correct, it's difficult to interpret because those spikes in the shadows due to the quantization make it difficult to find out how much information there is on every stop.
But there is a statistical trick to have a much more obvious (softer) histogram: averaging RAW data fills all levels in the histogram with the amount of light they received (or a very precise approximation of it thanks to statistics). The histogram I showed previously was a 8x8 pixels averaging. This allows to calculate scene's DR with cameras having less DR that the contrast of the scene. Actually the limit is only the RAW bitdepth.
Regards
¡Brillante!
-
A quickish 3 exposure hdr of the cat photo done with Lightroom and some local adjustments. With the camera used here, there is definitely a need for HDR to hold detail in the shadows and highlights and to keep shadow noise down. But the HDR process isn't the end of the story. You usually still have to massage the image to get what you want out of it. There is just the slightest bit of black clipping in the cat.
-
I bracket medium format film shots. I do that in case I screw up the measured exposure setting. My bracketing is 0, +1 and -1. Which programs could I try HDR that works with film? Any special things I should know? Can I use TIFF files? Thanks.
PS. I have Lightroom 6.0 purchased and PS Elements. Note that my LR has a plug-in called HDR Efex Pro 2. But I've never used it.
-
The RAW zone system reveals that the cat was not the darkest area: some parts of the furniture are even darker than the feline:
(each gray colour represents one stop, two highest and lowest stops are highlighted in other colours)
(http://guillermoluijk.com/misc/blackcatzones.png)
Regards
-
But there is a statistical trick to have a much more obvious (softer) histogram: averaging RAW data fills all levels in the histogram with the amount of light they received (or a very precise approximation of it thanks to statistics). The histogram I showed previously was a 8x8 pixels averaging. This allows to calculate scene's DR with cameras having less DR that the contrast of the scene. Actually the limit is only the RAW bitdepth.
Regards
Guillermo,
How did you made the averaging? Was that by using software you developed?
thanks
-
Guillermo,
How did you made the averaging? Was that by using software you developed?
I wouldn't call it software, it's just a suboptimum loop written in R:
# dcraw -v -r 1 1 1 1 -o 0 -S 16382 -4 -T blackcat.dng
img=readTIFF("blackcat.tiff", native=F, convert=F)
REDUCTION=8
DIMY=as.integer(dim(img)[1]/REDUCTION)
DIMX=as.integer(dim(img)[2]/REDUCTION)
imgresize=array(1,c(DIMY,DIMX,3))
for (i in 1:DIMY) {
for (j in 1:DIMX) {
for (k in 1:3) {
imgresize[i,j,k]=mean(img[((i-1)*REDUCTION+1):(i*REDUCTION), ((j-1)*REDUCTION+1):(j*REDUCTION), k])
}
}
}
writeTIFF(imgresize^(1/Gamma), paste0("imgresize_",REDUCTION,".tif"),
bits.per.sample=16, compression="LZW")
-
Thank you very much!
-
Attached is a rendering from the ETTR'ed dng (image 8 of the series). Raw conversion performed in RPP, with various adjustments done in PS, including some chroma noise reduction (Neat Image) and sharpening, color balance and local contrast. You can extract a serviceable image for the web, but if you were to print this you would probably want to blend/combine/merge another shot at a longer exposure for the shadow data. The camera focused on the curtain and the DOF was not sufficient to keep the cat in focus - you can see some small bokeh balls from the specular hits around the ears and whiskers. Those are easily removed with spot healing or similar touch up.
If you were going for a brighter rendering with a lot of boost to the shadow tones, like the cat, then it would be imperative to add another exposure.
kirk
-
Some Phase backs provide such raw Histogram as can some Canon's with:
https://www.magiclantern.fm/
However it's done under the hood, it's doable and sadly rare. Because for some strange reason, most manufacturers don't want to (not can't). 😩
As an aside...
When it comes to improvements thro' fireware updates, Sony and Hasselblad are way better than Canon. In fact, with Canon, apart from updates for new lenses and bug fixes, you pretty much have to wait for the next camera.
Thanks for the headsup on Magic Lantern. New to me.
D.
Edited. I see 5D MkIV not covered by Magic Lantern. Pity.
-
Interesting discussion...
Unless I missed it, no one here has discussed processing files using Linear (as Capture One call it) or Reproduction/Reproduction Low Gain (as Hasselblad call it).
Basically, you remove the baked-in curve, and whatever else the RAW processing software manufacturer considers a file from your camera model should look like straight out of the camera, and you start from scratch.
I don't think that ACR or LR have this option, but I'm open to correction.
I use it all the time for Sony, Canon and Hasselblad.
If you haven't tried it, you might be pleasantly surprised.
It is not for speedy processing, but then good processing is never fast.
D.
-
I don't think that ACR or LR have this option, but I'm open to correction.
http://www.color.org/scene-referred.xalter
-
http://www.color.org/scene-referred.xalter
I've used that method before and it seems to work. However, the linked article refers to Photoshop CS3, which is quite old.
I think it is necessary to use Process Ver 2.0 rather than the current PV 5.0. I presume it would be best to use the Flat camera profile if it is available with the camera used to capture the raw file.
Bill
-
http://www.color.org/scene-referred.xalter
I find Adobe Standard to be an adequate, relatively neutral starting point for my purposes. Sort of my digital "work" print.
-
Interesting discussion...
Unless I missed it, no one here has discussed processing files using Linear (as Capture One call it) or Reproduction/Reproduction Low Gain (as Hasselblad call it).
Basically, you remove the baked-in curve, and whatever else the RAW processing software manufacturer considers a file from your camera model should look like straight out of the camera, and you start from scratch.
Linear curve will not bring you a raw histogram. First, a curve does not affect the end points (0 & 255) and even more important, the image & histogram you see in Capture 1 is already color space encoded, which makes the histogram to look different.
There are many occasions where clipped channels are a result of out of gamut issues related to the specific color space, where the raw data is not clipped. Moreover, usually when one thinks out of gamut only considers the maximum values (255) , but there are cases when out of gamut is 0 because the mathematical value would be negative (i.e. the cyan patch in the colorchecker, third row to the right when using sRGB)
Another misconception: there is no "baked" curve in a raw file.
-
Linear curve will not bring you a raw histogram.
Somhow, yes, on a log scale. Even quite accurate with a repro profile and a linear scientific curve in capture One.
Another misconception: there is no "baked" curve in a raw file.
yes, not in the raw file, just to the default opening in most rawconverters. (ADobe, CaptureOne, Photolab, etc). After a fidelity base, a tone curve is most often applied on top. There is also some baked color processing (color-twist, memorycolors, etc) for more pleasing multi-purpose rendering -> avoiding mangenta skies, preserve skin tones from pink casts, green foliage, etc.
But ilf you are not interested in, for exemple packshoots photography, etc. Sometimes it's easyer to work on a base of color reproduction profiles. Moreover, working with linear profiles and a linear curve can be more suited for very precise works, especially in white shades, with a better control of the roll-off in highlights.
To Dinarius :
A linear curve with a roll-off can be a bit more suited for high dynamic range scenes (I think), but this kind of scene will also benefit from a local tonemapping, not only a bent on the tonecurve (wich does more or less in this case the same thing than pushing shadow/highlights recovery sliders.)
You can have linear values in ACR easily : process in v2, with everything at 0 including sliders and curves.
If you want the benefits of latest process, (highlights color recovery especially), you will need to make linear profile (lumariver profile designer, same dcamprof, basiccolor input), or remove a tone curve from an existing profile (you actually can do that easily with adobe dng profile editor). Then you will need to adjust the exposure with the white point from curves.
-
Somhow, yes, on a log scale. Even quite accurate with a repro profile and a linear scientific curve in capture One.
If this were true you would not need a profile at all. You would just use the raw data for repro. I bet it would not work.
-
If this were true you would not need a profile at all. You would just use the raw data for repro. I bet it would not work.
Somehow, it's a raw histogramm with WB adjustment. The repro profile (or a generic one) doesn't change much the histogram.
(https://i.ibb.co/Px3MfFz/histo.jpg)
-
I've used that method before and it seems to work. However, the linked article refers to Photoshop CS3, which is quite old.
I think it is necessary to use Process Ver 2.0 rather than the current PV 5.0. I presume it would be best to use the Flat camera profile if it is available with the camera used to capture the raw file.
Bill
Adobe supplied DCP profiles use non linear color transforms, hence using them and calling the result (output after ACR/LR) "scene-referred" is plain stupid
-
Adobe supplied DCP profiles use non linear color transforms, hence using them and calling the result (output after ACR/LR) "scene-referred" is plain stupid
Right on, what does the ICC know about color when we have a color expert like you telling us the colorimetric 'facts'. ;D
http://www.color.org/scene-referred.xalter
-
Not so fast Andrew ;) . DP mentions a "scene-referred" profile. To me this sounds like a kind of an ‘adaptive profile’, a profile that adjusts itself based on the actual content of the scene on hand. We all know this does not exist and as such I fully agree with DP on Adobe’s “scene-referred” statement.
Care to share your expert opinion on this?
Cheers,
Jaap
-
Not so fast Andrew ;) . DP mentions a "scene-referred" profile. To me this sounds like a kind of an ‘adaptive profile’, a profile that adjusts itself based on the actual content of the scene on hand. We all know this does not exist and as such I fully agree with DP on Adobe’s “scene-referred” statement.
Care to share your expert opinion on this?
Cheers,
Jaap
Yes: REREAD his wrong statement.
hence using them and calling the result (output after ACR/LR) "scene-referred" is plain stupid
Now read what those ”fools” at the ICC describe as the results. 😂
“There are two ways to be fooled. One is to believe what isn't true; the other is to refuse to believe what is true.” -Søren Kierkegaard
-
If this were true you would not need a profile at all. You would just use the raw data for repro. I bet it would not work.
You can try it, and you get "something" visible. This is what RAW data assigned to sRGB looks like prior to WB:
(https://1.bp.blogspot.com/-CDhaPYW4xJg/WxQi3oTpO3I/AAAAAAAADvE/JMgOBUhTu1wPMNhMB86OU3tXHf8C4E9-wCLcBGAs/s1600/namibiarawrgb.png)
Composing every pixel colour from 4 RAW pixels: {R, (G1+G2)/2, B}, you get:
(http://www.guillermoluijk.com/datosimagensonido/debayer.png)
and here after WB (performed through a linear scaling on R and B), but with no colour profile conversion applied yet (RAW white balanced RAW values are directly assigned to sRGB):
(http://www.guillermoluijk.com/datosimagensonido/debayerwb.png)
The image looks pretty decent, and its histogram is a good proxy for the RAW histogram. The bad news is that no commercial RAW developer nor camera provides something like any of those two images. All of them apply white balance and colour profile conversion (which mixes up the three RGB components), so having a "Linear" option in your camera or RAW developer is not very useful if you are interested in the genuine RAW histogram.
Regards
PS: processed image with proper colour profile conversion:
(https://3.bp.blogspot.com/-_P2l-zqYaL8/WxQbLnDWSJI/AAAAAAAADuk/A-KKNMGk1d8v1kZEsolRLI-Uw4_k2rgogCLcBGAs/s1600/namibia.jpg)
-
Linear curve will not bring you a raw histogram. First, a curve does not affect the end points (0 & 255) and even more important, the image & histogram you see in Capture 1 is already color space encoded, which makes the histogram to look different.
There are many occasions where clipped channels are a result of out of gamut issues related to the specific color space, where the raw data is not clipped. Moreover, usually when one thinks out of gamut only considers the maximum values (255) , but there are cases when out of gamut is 0 because the mathematical value would be negative (i.e. the cyan patch in the colorchecker, third row to the right when using sRGB)
If you want a histogram of the image in a linear space, you could first render into ProphotoRGB and then convert to LinearRIMM Ver 4 as per the ICC article that Andrew cited.
Bill
-
Yes: REREAD his wrong statement
my correct statement was "Adobe supplied DCP profiles use non linear color transforms, hence using them and calling the result (output after ACR/LR) "scene-referred" is plain stupid"
PS: if you refer to the quite ancient articles @ color org from the times when Adobe supplied only matrix profiles then you at least need to have some decency to add notes about that (matrix vs what kind of DCP profiles will be used NOW unless you take care not to use them) ;D
-
my correct statement was "Adobe supplied DCP profiles use non linear color transforms, hence using them and calling the result (output after ACR/LR) "scene-referred" is plain stupid"
PS: if you refer to the quite ancient articles @ color org from the times when Adobe supplied only matrix profiles then you at least need to have some decency to add notes about that ;D
We are not and should not drink your bleach. The results of the raw as the color scientists at the ICC describe are clear. And obtainable.
-
....
The image looks pretty decent, and its histogram is a good proxy for the RAW histogram. The bad news is that no commercial RAW developer nor camera provides something like any of those two images. All of them apply white balance and colour profile conversion (which mixes up the three RGB components), so having a "Linear" option in your camera or RAW developer is not very useful if you are interested in the genuine RAW histogram.
Regards
...
Raw Photo Processor provides settings which will permit you to produce an output file identical to:
dcraw -T -4 -r 1 1 1 1 -o 0 <raw filename>
i.e., linear, 16bit TIFF, UniWB, raw color output. With these settings in the application, the histogram will display the raw histogram that you would get if you viewed the raw file in Raw Digger or FRV. Pretty handy.
Kirk
-
I know Kirk, but RPP is not a RAW developer any photographer would use to process his/her real work, it's just a DCRAW front end like some others have existed for a while. I was talking about mainstream commercial RAW developers like ACR, LR, C1,... or any camera. Much closer to the concept of a RAW developer with RAW histogram abilities is RAW Therapee, but again a non-commercial niche tool.
Regards
-
I know Kirk, but RPP is not a RAW developer any photographer would use to process his/her real work, it's just a DCRAW front end like some others have existed for a while. I was talking about mainstream commercial RAW developers like ACR, LR, C1,... or any camera. Much closer to the concept of a RAW developer with RAW histogram abilities is RAW Therapee, but again a non-commercial niche tool.
Regards
It is still puzzling why all camera manufacturers offer raw file acquisition, but none offer a set of tools for evaluating the acquisition of raw files. FWIW, I use RPP for my "real" work, but I do not consider myself a photographer, so I guess your point stands!
Also, my XPro-3 and my X100V both shoot in HDR mode where three separate exposures are acquired and blended, in camera, to produce a single raw output file. It's almost like having Zero Noise built in to my camera, except from my initial evaluation of the files, the exposure increments are not large enough to rival the output of your application. Still, better than nothing.
Kirk
-
It is still puzzling why all camera manufacturers offer raw file acquisition, but none offer a set of tools for evaluating the acquisition of raw files.
We hoped this for years. I consider it's time to give up waiting for such a feature: camera manufacturers clearly didn't consider it a plus, and it will become less and less necessary as digital sensors improve their performance and image processing relays more and more on IA. RAW histograms or clipping warnings are nice as an analysis feature, but a photographer is not going to analyse his pictures but process them. So having the RAW histogram when you cannot use it to improve capture becomes of little use to the photographer. It's rather a forensic tool.
Regards
-
We hoped this for years. I consider it's time to give up waiting for such a feature: camera manufacturers clearly didn't consider it a plus, and it will become less and less necessary as digital sensors improve their performance and image processing relays more and more on IA. RAW histograms or clipping warnings are nice as an analysis feature, but a photographer is not going to analyse his pictures but process them. So having the RAW histogram when you cannot use it to improve capture becomes of little use to the photographer. It's rather a forensic tool.
Regards
I started using my P&S as a light meter when I shoot film. Would having RAW histograms make the P&S a better meter?
-
I started using my P&S as a light meter when I shoot film. Would having RAW histograms make the P&S a better meter?
If you are using the meter ideally, for the raw data, then no. Again, some of us optimally exposed film (and tricky transparency) without a Histogram, for decades.
-
If you are using the meter ideally, for the raw data, then no. Again, some of us optimally exposed film (and tricky transparency) without a Histogram, for decades.
I've been using a Minolta light meter for years as well. But with medium format, I'd bracket because film is cheap relatively. But I just started using a 4x5 camera, all new to me.
I decided to also try using a digital P&S camera as a tool as well. It allows me to line up a 4x5 shot before setting up the LF camera, setting the zoom to match the view angle of the 4x5 lens, and to see what the picture looks like in BW when I'm shooting BW film. Of course, I'm also trying to use the camera as a meter as well. Beside histograms, I can use it as a spot meter or center meter. It will get more interesting when I start shooting Velvia 50 in 4x5. I'm saving this film for when I have more experience shooting a large format camera, a challenge in itself.
If you were going to use a P&S camera and take advantage of its histogram, and spot and center metering, what would you be looking for when shooting chromes and BW negative film?
-
I started using my P&S as a light meter when I shoot film. Would having RAW histograms make the P&S a better meter?
RAW means linear light measurement, and this means the ability to calculate the precise relations of light in stops in the scene. In the end each individual pixel in a digital sensor acts as a light meter, and you have millions of those. Perhaps you can get some cheap Canon DSLR model working with Magic Lantern; this firmware allows for RAW histograms expressed in EV stops:
(https://www.magiclantern.fm/forum/proxy.php?request=https%3A%2F%2Fsnag.gy%2FMEWRy0.jpg&hash=7e9b722609de318469a7f687bd655456)
But Magic Lantern is not an easy path to install and have it working.
Regards