HDR vs. Compositing

shmne

No longer a newbie, moving up!
Joined
Jun 3, 2009
Messages
641
Reaction score
83
Location
Florida
Can others edit my Photos
Photos OK to edit
Ever since the term "HDR" was introduced to people it seems its actual meaning has been greatly confused by the masses. So I am going to attempt to make a post that clearly explains this concept, versus another concept called compositing.

The following is an article I've decided to write to clarify HDR, and compositing, and when to use each.

High Dynamic Range (HDR) - A technique in which the end goal is to allow the viewer to see a broader range of lights to darks without losing information due to clipping.

What this means: When taking a picture, mainly outdoors, it is sometimes impossible to collect all the light information in a scene. Sometimes, you can only gather soo much information before you have to start cropping either the highlights or shadows. So to avoid the loss of information you take multiple exposures, usually 3, so that later you can use the photo-manipulating software of choice to composite them into one congruent piece.

How it works: By having a more highlight values (or lights), and more shadow values (or darks) you have effectively increased the dynamic range of the image. Dynamic range is literally the ratio between these two values.

Your computer actually takes the separate images and combines them in a way to increase this ratio drastically, by using an algorithm. This algorithm is designed in a way such that more details are created in your lights, and in your darks. So now you have more information in your lights, darks, and even grays which in turn makes a more appealing shot because our eyes love information.

Compositing - (Taken from web definitions) is the combining of visual elements from separate sources into single images, often to create the illusion that all those elements are from one originally.

So you might be finding yourself asking a question right about now, something like;
"But say, Joe, if an HDR image is just combining multiple images to look like one, isn't HDR a form of compositing???"
And I would have to respond with yes, in fact HDR is compositing.

What this means:
This means when you are taking a shot of your nifty shoes, and use HDR techniques, you are not creating an HDR. In fact your are simply adding saturation, contrast, a fake level of sharpening, and generally speaking an increase in the dynamic range of colors.

Bringing it all together now: Taking the steps to produce an "HDR" image when it does not require it, does not actually make the image HDR. This is because you are not increasing the dynamic range of light at all. At most you are increasing the dynamic range of colors, which could be done much simpler by duplicating the image and throwing it in overlay with photoshop.

All HDR is compositing, but not all compositing is called HDR. The sooner you can understand this, the sooner you can take advantage of using compositing techniques that do a better job then HDR does. HDR is a great tool, but like all tools it is important to use it at the right time. There are many techniques out there that do a better job making a photo look "pro"

I hope this clears some questions up, and more importantly I hope this makes you ask yourself the next time you are out shooting "Do I really need to make this HDR?"

P.S. - Sorry if anything is a bit off, it is currently 7:40 a.m. and I have been up all night. I have felt the need to write this up for some time now and I hope I did a decent enough job explaining the two concepts clearly. Any critiques are appreciated to make this more streamlined and understandable to people.
 
Good definitions

So, as an example - say I take a picture of a truck in direct sunlight and in order to properly expose the focal point of the picture I have blown out highlights in some areas and shadows in others that are under exposed. Now bear with me...then I take the tools in Lightroom and adjust it to bring out shadow detail and pull details back into the blown out areas...

"HDR", right? Since the photo was impossible to properly expose corner to corner...

...isn't the finished product displaying a higher dynamic range than what was possible out of a single shot under the conditions in which it was shot?

So, what do you think?

-not being a smart ass, but I have a degree in medicine - I can grasp basic physical concepts.
 
Shmne a well written explanation. The only thing that might be made a little clearer is about describing when somebody takes the boring image of their sneakers then cranks the contrast and saturation to make a more pleasing image. They have increased the dynamic range of *their image* but they have not increased the dynamic range of the scene they captured. I don't even like how I just said it, but I think you get the picture. HDR means capturing a scene that has way too much dynamic range to be captured in one exposure so you use compositing to combine images that span across that large dynamic range.

Modlife: I think the concensus on HDR is that it's from multiple exposures. Your example inferred that you only took 1 exposure. This leads me to my final part:

And now I have a question for you, shmne: Have you ever taken an image where the histogram shows lots of info in the shadows, lots of info in the highlights, and basically nothing in the midtones (for example a landscape shot with sky in the background)? If you were to create layers and do a contrast stretch of both areas seperately and then composit them back together, is this not the same thing as HDR just without the extra exposure? I think the answer is "no" because I didn't actually record a wider dynamic range scene. Even though we're doing the same type of compositing that HDR does, I suppose it's not officially HDR.

That then brings me to our discussion on the Beyond The Basics forum. If a RAW records quite a larger dynamic range than any JPG derived from that RAW, can you not create 2 distinct "exposures" (actually JPGs) from that 1 RAW? The first "exposure" would show details in the shadows and clip the highlights. The second "exposure" would show detail in the highlights and clip the shadows? You could then use these 2 distinct JPGs just like an HDR guy would use 2 distinct exposures. What do you think?
 
That then brings me to our discussion on the Beyond The Basics forum. If a RAW records quite a larger dynamic range than any JPG derived from that RAW, can you not create 2 distinct "exposures" (actually JPGs) from that 1 RAW? The first "exposure" would show details in the shadows and clip the highlights. The second "exposure" would show detail in the highlights and clip the shadows? You could then use these 2 distinct JPGs just like an HDR guy would use 2 distinct exposures. What do you think?

I know at this point you guys nay not want to listen to me, but this is exactly the method that many "popular" HDR advertisements and shots were made with.

The "proper" way to make a composite HDR from a single exposure is to make copies of the original RAW - with exposures bumped equally up and down (to end with -1,0,1 as an example. After that you can absolutely use photoshop, photomatix, or whatever to create the composite and tone map.

With one exposure you can't have the same amount of usable data, but I'm convinced that due to the nature of RAW, you can create HDR from one file - the RAW file format captures a range of data that is wider than what you can see at any give exposure.

If you have a moving subject (or are moving yourself) then you currently have no choice but to use one exposure - the technology isn't there.

I'll argue for years about this - one RAW file has more data than what you can see, so HDR is possible....

...the greater number of images over the widest range of exposures will give you the best quality in a composite image

Like I said before - try some of the same composite methods with jpegs and raw files and you'll see that Raw files actually do have tons of extra, and useful data in them.

An HDR from 20+ exposures of JPEG files would have a hard time competing with the same RAW files over a much tighter range with fewer exposures.

Now, I'm not an expert at techniques - that's an art of it's own. I do understand technology though and I'm certain of what I say.
 
But real, good HDR, doesn't make an image look crappy.
 
@Modlife

What I was calling you out on was pushing the sliders to 100% and calling that HDR. That is ruining image quality, because those sliders are like effects being applied to an image, and there was plenty of evidence in your shadows to prove I was right. Also, again, after posting it so many times, anything to be considered high dynamic range needs to have originally had a dynamic range not being able to be captured with a 12 bit sensor, or film for that matter.

That link you put up also confirms what I said, look at his shot. Also thank you for posting it because it helps me for stosh at the end and confirm even more so that a single exposure HDR is not as good as a true HDR.


@Stosh

Yes, more or less. You can create HDRs with RAW files because of the extra captured range, however I actually just answered this in that same thread. When we were created an HDR with a single exposure, then compared it to a proper HDR the histograms show a great difference in dynamic range. There is just more information in a proper HDR. Is it really that much better? Heck no, not if you can't get more exposures easily. And btw, it was the neutral shot from the set of 3.

Could there have been human error that could have accounted for this? Of course, I'm just stating the point of the one lab activity.

And sorry that wasn't an attack on those shoes xD I feel bad now! I just meant that there wasn't a reason to create an HDR, and that was the first immediate shot that came to mind when I wrote this.

When you have sky (which will almost always overexpose and CLIP = no more data) and darks (which will also CLIP = no more data) a single exposure will not be effective; even in RAW!!. It will be close, but not good enough. This is why I get annoyed when people use single exposure HDRs in wrong situations, because you aren't expanding the light to darks, I mean yes slightly you may increase the dynamic range, but instead you are expanding tonal dynamic range which is a completely different beast! A 12 bit file can only capture soo much dynamic range before clipping one way or another will occur, just like an 8bit image. If any one "Pixel bucket" gets over filled then you are now losing that "pixel bucket" and the others near it as well!

To make a proper HDR you must convert to 16 bit :) Then proceed to use those .PSDs to make the HDR image. At least that is the way you really should do it, because otherwise you can not expand on dynamic range! You are limited by what the RAW had in the first place, thus making the point null and void. When you scale down now, to 12 bit, then to 8 bit, you actually retain more information because you started with more. Yes yes yes, at the end you still only have 0-256, but trust me there is more information present. Or don't trust me and do it on your own! It is this concept that allows a proper HDR to include a better dynamic range then something created from just one exposure. Could you do the same with one? Yes, as I said earlier, will it look good? Oh yeah! But will it be just as good as a true HDR? No way! Should you always do a multiple exposure HDR? NO WAY!! You should use the tool that is apprioate. Always, end of story. Don't try to shove a pentagon into a square, is my motto :)

Sorry this is a bit wordy at times, I'm also goig back to clear up some of my original post! And probably explain the process fully so people can see the steps 1-fruition of a true HDR, as well as a single exposure HDR. Only if I have that kind of time though xD That is a lot of writing!
 
Good definitions

So, as an example - say I take a picture of a truck in direct sunlight and in order to properly expose the focal point of the picture I have blown out highlights in some areas and shadows in others that are under exposed. Now bear with me...then I take the tools in Lightroom and adjust it to bring out shadow detail and pull details back into the blown out areas...

Wrong, if you clipped your sensor there is no pulling back the blown areas, and if there are underexposed parts that clipped then there is no pulling it back from death.

If you clip, you clip, end of story. You are not creating a higher dynamic range, you are restoring the original dynamic range.
 
An HDR from 20+ exposures of JPEG files would have a hard time competing with the same RAW files over a much tighter range with fewer exposures.

Really? 20 exposures of JPEGs couldn't have the same dynamic range period, no matter how hard you tried. You would have to convert them to a 12 bit depth file, then you could. Otherwise you are stuck in an 8bit world so this point is completely moot.

And yes, most "HDR" shots are made with one shot, I never said they weren't. It just isn't a true HDR because you are not expanding on the dynamic range -_-. The camera sensor still only picked up on its fullest range and no more.

But I think you are missing the point of an HDR in the first place, it was so that the blown out areas wouldn't be blown out, and the underexposed areas wouldn't be underexposed. If you can take a picture of something without that happening, then there is no reason to call it HDR because it just simply isn't. At that point you are creating a composite in order to add extra detail, saturation, contrast, etc. Yes the whites can get whiter, and the darks can get darker, but if this is all you wanted to do in the first place then it wasn't needed to make an HDR at all -_-
 
Last edited:
I think you guys needs to google the difference between HDR and Tone mapping. you can't HDR a single image but you can tone map a single image. I've been in HDR mode for a little over a year and have sold quite a few HDR items, Have always used 3 images to produce quality HDR photo's. Never have and prolly never will try to pass off a single image as a hdr photo, have tried doing a single photo in photoshop and also in camera and it just doesn't look the same as a genuine produced hdr after being run through photomatix. So the theory of producing a hdr photo from a single image i'm afraid to say in my opinion is pretty much hogwash
 
To us HDR is a process. To a consumer or customer isn't HDR more of a look? If you use a single exposure to create the look they want, what's the difference? Would they be disappointed to later find out that you "passed off" a single image exposure to be an HDR?
 
Good definitions

So, as an example - say I take a picture of a truck in direct sunlight and in order to properly expose the focal point of the picture I have blown out highlights in some areas and shadows in others that are under exposed. Now bear with me...then I take the tools in Lightroom and adjust it to bring out shadow detail and pull details back into the blown out areas...

Wrong, if you clipped your sensor there is no pulling back the blown areas, and if there are underexposed parts that clipped then there is no pulling it back from death.

If you clip, you clip, end of story. You are not creating a higher dynamic range, you are restoring the original dynamic range.

But it is possible to shoot a RAW file and what appears to be blown out area will still have viewable data as long as you drop the exposure in post and didn't totally over expose that portion of the photo in the original exposure.

Shooting to the right.

And various articles have stated that when shooting to the right and recovering the apparently blown highlights, you can created a slightly greater dynamic range in the photo. You're not going to get three whole stops of dynamic range of course.
 

Most reactions

New Topics

Back
Top