Beginner's question - how to do HDR?

[nikond80hdr]

Hi people. I own a Nikon D80 and I wanted to ask; is it easy to do HDR images?. I have a good enough grasp on CS3, but I am still learning photography and the whole HDR thing seems wonderful, if not a little confusing.

It's probably that HDR itself is pretty simple, but the people I have heard online trying to explain it, seem to either include EVERY possibility and margin for error, and make it seem daunting, OR they expect you to know most of it, and miss things out. Then, you have the guys who are just appalling at writing tutorials (like me maybe).

Can someone explain, from start to end (or link me to a good guide) how I begin with HDRI?.

Thankyou folks.

 

[nickoly]

I'm a beginner too. You may find that CS3 is all the software you need. I have used Photomatix. There is a free version of this and the tutorials therein are probably enough to get you going.

Here is an example, where I got caught short by the lighting contrast and used Photomatix to salvage a result, using a single shot in RAW.

http://forums.dpreview.com/forums/read.asp?forum=1008&message=25759810

You will also find some good comment and examples on other threads in the lighting forum.

http://forums.dpreview.com/forums/read.asp?forum=1025&message=25975081

You can safely ignore any posts from the starter, Walker, he just a pathetic fantasist who has nothing to show.

I submit that one of the most vital steps on the learning curve is the exercise of restraint. Don't let anybody kid you that HDR post processing is the answer to everything. It's just another tool. If you want surreal landscapes, or rather strange interiors that may be well suited for real estate ads but nothing else, then fine, but you will find the word "awesome" lavished on this sort of work even more often than it usually is around here. What they really mean is "awesomely cr@ppy" so just be careful!

 

[MrRoger]

HDR stands for High Dynamic Range. It is a very general term and may be describing different techniques and applications.

The two most common uses in photography are:

1) Enhancing the tones in an image so that you can appear to display a greater range of tones (from dark to light) than would be possible on the output media using the correct tone mapping.

2) Merging 1 or more photograph at different exposures in order to overcome the limited dynamic range of digital cameras.

Of course, in some situations you could be doing a combination of the two.

I know little about 1) in situations were it involves mapping a single image.

Regarding number 2, consider this. You have a landscape scene with a wonderful sky with fluffy clouds and a forground partly in shade.

If you expose for the sky, the clouds will look wonderful, but you will have not have any detal (just black areas) in the shade. Conversly, if you expose for the shady areas, your sky will be blown out.

So you take two pictures, one exposed for the sky and one exposed for the shade. Usually you try to arrange this such that you only change the shutter speed and keep the lens aperture the same, you could also vary ISO on your camera as noise difference is minimal between 100 and 400 ISO.

Ideally a tripod should be used, but this is not essential with modern software as it can do a pretty good job of aligning images as long as they are very similar. This is way bracketing can be useful, you don't need to change anything between shots, just hold the camera steady, but you can always practise changing exposure without taking your eye of the viewfinder, or use reference marks in the image (what is in the centre ) so you are sure that the images are fairly well aligned.

Shooting RAW is better, but AFAIK, CS3 does not support merging RAW images.

The you merge the two photos together. This used to be very tricky, and it may be that some of the complex descriptions you have run into regard people who are still using older techniques.

I use PSP X2, CS3 has very similar features, but more of that later. Using HDR photo merge on the file menu you load the images you want to use. You can use as many as you like but I would start by using 2 or 3, inexperienced use of two many images will make things worse. Click the 16 bit option to get a 16 bit image out.

Once the images are loaded you can hit the align button, essential if they were not tripod shots. There are settings for brightness and clarify, I would suggest that you use the suggest settings button, this will get you in the ball park and you can fine tune later.

Once you have your merged 16 bit image you will find it appears to have quite low contrast, some PP work is generally essential to emphasize the photo the way you want it. If you have used RAW then you should probably adjust the WB first, if necessary. To get the right contrast, if you find curves tricky, then I suggest you use levels first, then highlights/midtones/shadows and possibly clarify, followed by normal brightness contrast to get the final image.

CS3 does pretty much the same, except that it does not do RAW merges, the alignment does not appear to be so good, it first makes a 32 bit image which then has to be downgraded to 16 bit to actually process, and many people say that the prgram often crashes (Personaly this has only happened to me a couple of times, and allthougth I have not used it much, my experience suggests that it is useable).

By the way, there is no reason why you can't use both, PSP is quite cheap and having merged the photos you can save it in PS format and transfer, that way you get the best of both worlds (I am assuming you prefer CS3 for post processing, you could just do everything in PSP which is what I do).

That should get you started. Start with an easy task, like the landscape with an exposure for ground and sky, then progress.

 

[nickoly]

Shooting RAW is better, but AFAIK, CS3 does not support merging RAW
images.

I would not expect any HDR process to support RAW. RAW is merely the source to give you the files of varying exposure, but allows you to get them by taking only one shot.

 

[MrRoger]

Ehhh?

HDR photo merge, in both CS3 and PSP X2, allow you to load 1 or more image at different exposures and merge them to get an image with higher dynamic range.

Obviously if you use RAW rather than jpeg you get higher resolution plus more potential range in the original image.

Why not use RAW, it works better

 

[nickoly]

Ehhh?

HDR photo merge, in both CS3 and PSP X2, allow you to load 1 or more
image at different exposures and merge them to get an image with
higher dynamic range.

I understand that, but I don't understand how they would use a RAW file. The only way I have used RAW is as a source for four different files, all BMPs, of different exposure settings.

Obviously if you use RAW rather than jpeg you get higher resolution
plus more potential range in the original image.

That is not obvious at all. I don't think RAW offers any higher resolution, just RAW data for post processing. Yes, there is more potential range.

Why not use RAW, it works better

So, have you actually used RAW files in merge process? This completely baffles me.

 

[GaryDeM]

to do hdr-

if you put camera into full auto matrix metering, take first shot note fstop and shutter speed. put camera into full manual, see if camera still has the matrix fstop and shutter speed. if yes, then using shutter speed go up 2-4 shutter speeds 1 fstop worth of shutter speed at a time. the back to matrix shutter speed and go down same number of shutter speeds.
this is on a tripod with cable release.

no, you should not use 1 raw shot and convert 1 stop up and down, because their is not enough dynamic range in the 1 raw shot. dynamic range is why we are doing this, hdr is trying to get all it can.

the group of shots can be raw or jpeg. if jpeg they can be used as is. if raw remember that you HAVE to batch process all 5-9 shots. this is because the pp has to be all the same on every pic. you cannot, for example make any attempt to get the shadow details of the group of raw pics, because that would require different amounts of pp, and you cannot do that with hdr. the pp for all shots has to be identical.

for me i just shoot them in jpeg and use them from the camera, that way they are all identical because the camera jpeg settings are the same for every shot. i also put my hand streched in front of the lens and take a check shot and when done take a ending shot with hand. this tells me where the hdr group is on my memory card when i transfer to the pc.

the only important item is to bracket using shutter speeds only. if fstops are used it changes dof between shots. and shoot enough shots, 5-9 is the optimum. the only other thought is to shoot a scene that deserves the the hdr technique, too many people are shooting hdr because it is new or different or whatever. many people are using hdr software on scenes that do not have enough dynamic range; they end up with images that have been enhanced by hdr software, they are not hdr images. the dynamic range was not in the scene to begin with. the scene for hdr should have a very wide dynamic range. this can be checked with a spotmeter on different areas. NOTE: use of auto bracketing on a camera may not work unless you know the bracketing is using the shutter speeds to bracket. in any event, you really need 5-9 shots for hdr; this is more than the auto bracket fcn on almost all cameras. and the bracketing has to be both sides of the middle shot. make you use enough brackets to cover the previously checked dynamic range. it does little good to bracket for a 10stop dynamic range when the scene has 14stops.

and the scene should have no movement, if so the item will blur in the hdr image.

do not adjust the focus. set the focus on infinity or use a hyperfocalsetup for focus.

do not adjust the white balance for individual shots. go with awb or 1 setting and do not change it.

remember, hdr was created and meant for scene that have a dynamic range that exceeds the dynamic range of the camera sensor, about 5-6stops. hdr with the required software allows the user to capture a scene that has very high dynamic range.

 

[nikond80hdr]

Good post; lots of wisdom. Thanks!.

 

[MrRoger]

The format of RAW files varies, but it is nothing more than an unprocessed image from the sensor that contains more detail and info than a processed jpeg, but requires PP on your computer to adjust the image.

If you shoot RAW you will eventually need to PP the image for WB, NR, and shifting the levels to get the required contrast range. But you can do that before or after an HDR photo merge.

Doing the merge as the first step (i.e. merging the RAW images to a 16 bit result) is quicker, easier and better for NR as HDR merges tend to self cancel noise to a limited degree.

But don't take my word for it, go to Corel, download a PDP X2 trial and try it. It works and it works well.

The image below shows PSP X2 at the point where 4 Pentax RAW files are being opened for an HDR photo merge. Following the merge the image will be loaded in PSP as a 16 bit image ready for NR, WB curves operation etc.

 

[MrRoger]

if raw remember that you HAVE to batch process all 5-9 shots.

Or simply use a program that can merge the RAW images and let you process a single 16 bit result ;-)

the only important item is to bracket using shutter speeds only. if

You can use ISO as well, many cameras, including the OP, get no noticable noise increase going from 100 too 200, and maybe even 400. This may certinally be of interest for handheld shots.

NOTE: use of auto
bracketing on a camera may not work unless you know the bracketing is
using the shutter speeds to bracket.

On my camera, and I suspect most others, if you use aperture priority (A mode) exposure then bracketing will be done on speed.

5-9 shots for hdr; this is more than the auto bracket fcn on almost
all cameras.

This is certainly not my experience. I often take bracketed shots with 5 exposures at -2,-1,0,1,2 EV. I find that if I have spot metered correctly I usually get better results using just the -2, 0, and +2 than all 5. In fact given that there is generally at least 6 EV in each shot, there would seem to be a lot of overlap using so many exposures ;-)

Also, I often use just 2, one for land and 1 for sky. In these cases I will manually shift the camera down 3 steps for a cloudy sky, more for a clear.

 

[GaryDeM]

"This is certainly not my experience. I often take bracketed shots with 5 exposures at -2,-1,0,1,2 EV. I find that if I have spot metered correctly I usually get better results using just the -2, 0, and +2 than all 5. In fact given that there is generally at least 6 EV in each shot, there would seem to be a lot of overlap using so many exposures ;-)

Also, I often use just 2, one for land and 1 for sky. In these cases I will manually shift the camera down 3 steps for a cloudy sky, more for a clear."

the 5 shot bracket stated above will work PROVIDED that it covers the dynamic range of the scene. it does little good to bracket for 10stops if the scene is 14stops. for that reason i in my how to stated a range of bracketing. it being dependent on using a spotmeter on the dardest and lightest areas of the scene to determine the actual dynamic range that you have to cover;THEN decide on the number of brackets. which is why i suggested about doing the bracketing manually, then you can use any number of brackets.

if you use 2 shots, one the land and one for the sky that is not bracketing for hdr. it is not an hdr image. it blending of 2 shots into one.
see-
http://www.luminous-landscape.com/tutorials/blended_exposures.shtml
or you can get software to automate blending using DRI Pro, try google.

 

[MrRoger]

if you use 2 shots, one the land and one for the sky that is not
bracketing for hdr. it is not an hdr image. it blending of 2 shots
into one.
see-
http://www.luminous-landscape.com/tutorials/blended_exposures.shtml
or you can get software to automate blending using DRI Pro, try google.

Ok, I see I long winded procedure for "blending of two shots".

But I do this automaticaly using an automated procedure on the file menu called "HDR photo merge", why on earth should I call it something else and so it all by hand? The align images bit would be very tricky by hand.

HDR stands for High Dynamic Range, and by merging more than one image I do, in effect, get more dynamic range. What advantage would I get by doing all that manual procedure?

 

[MrRoger]

or you can get software to automate blending using DRI Pro, try google.

I am quite happy with with the built in "HDR photo merge" function built into my digital darkroom. Maybe it is heresay to do what I am doing....but it works!

 

[GaryDeM]

to your question-answer: i have no idea what you gain by doing all that manual procedure. i was showing that the use of 2 images is bleanding not hdr. the fact that you can use a hr software to get 2 shots into 1 does not make the final image hdr; it is still blending.

in any event i would use the suggested software. i have it, it works. it turns 2 images into 1. even gives instructions on the spread of the stops and hot to shoot what the program needs.

 

[MrRoger]

Is this because HDR is a pretty general term that some people use to refer to tone mapping which allows optimisation of tones within a limited color space, whilst others use it to refer to the fusion of 2 or more images with a limited but overlapping DR to yield a result with more DR?

Frankly, I cannot understand the basis of your diffrentiation. You say that combing two or more images at different exposures using an HDR photo merge function in order to capture a scene with a large DR is not HDR but blending.

OK, so what would consitute HDR? Do we have to exceed a certain EV threshold? Do we have to be shoehorning are larger DR into a smaller colour space? And given that all the above examples could occur with just 2 exposures, why must there be at least 5 exposures before we call it HDR photo merge?

Personally I have tried HDR merge with up to 12 images, but I find that seperations of less than 3 EV between exposures are counter productive, there is no tonal advantage but you can lose sharpness. You may gain passive noise cancellation but of course only if you were using high ISO images.

 

[MrRoger]

in any event i would use the suggested software. i have it, it works.
it turns 2 images into 1. even gives instructions on the spread of
the stops and hot to shoot what the program needs.

I have looked at this. It seems to be doing a limited subset of what I do with the HDR photo merge. I can also use 3, or n images. ANd the aligment works better.

I have posted this example before on this forum, but as the before and afters are set up, here goes again.......a hand held shot were 3 images (-2, 0, +2) form a 5 shot bracketed set have been automaticaly aligned and merged to get a 16-bit image.

Do we call this 3 shot belnding or a 3 shot HDR photo merge (as the makers of both photoshop and paint shop pro, the two largest DD software manufacturers) would suggest we call it.

+2EV



0EV



-2EV



Combing the images (this shows all 5 shots in the set, after using "align images" I try deselecting the intermediate -1/+1 and it almost invariably yields a slightly sharper image).



Final image after merging and a bit of PP with perspective correction:



All the processing required to get that final image can be done entirley within CS3 or PSP X2 using simple built in functions. Whilst I have no doubt that there may be some extra special fuctionality for specialists that can only be obataining using external apps or plug-ins, I think this is a good starting point for beginners.

 

[PicOne]

if raw remember that you HAVE to batch process all 5-9 shots.

Or simply use a program that can merge the RAW images and let you
process a single 16 bit result ;-)

With PSPX2, can you adjust each raw for eg. WB which setting is retained when you then send each raw into the merge?

the only important item is to bracket using shutter speeds only. if

You can use ISO as well, many cameras, including the OP, get no
noticable noise increase going from 100 too 200, and maybe even 400.
This may certinally be of interest for handheld shots.

By implication, ISO shifting only then allows you a 3-shot, 2-stop range for HDR merge, ie. 1 stop under and over 200 ISO (= 100 and 400 ISO). Not sure usually worth the effort to gain just 2 stops of DR.

--
'Everything in photography boils down to what's sharp and what's fuzzy.'
-Gaylord Herron

 

[GaryDeM]

my reply to the points you raised-

-hdr is not blending and vice versa. in blending you are simply using 2 shots separated by 3 stops(the dri pro software wants the user to take 1 shot 1 1/2 stops above the metered shot and the other 1 1/2 stops below the metered shot) this is simply a way of extending the dynamic rnge of the digital sensor. it is making no attempt to cover the dr of the scene. instead the effort is to get it more lifelike.

hdr is the attempt to cover the actual dr of a scene. many people are simply shooting, as you you said 3 or 5 shots, and saying that is hdr. it is, to the extent that it is greater than the normal shot.

• but the question is the 3 or 5 shot a true hdr image. to me there is no way of answering the question without knowing the dr of the original scene. that is why i suggested to spotmeter the scene's lightest and darkest areas to determine the actual dr that the user has to cover with his hdr image. for this reason i shoot 9shot hdr sets this gives me a much better chance of covering the dr of the scene. the 9 are separated by 1 stop each. if the scene needs more than 9 then i would do more.

as a photographer we are in the effort of recording on an image medium the scene that we are looking at as real and accurately as we can. on wide dr scenes hdr gives us the tool to do that with. i am not out there using hdr for the exercise, i want to get the best copy of the scene that i can. i do not consider a 3 or 5 shot hdr the proper way to do that unless the 3 or 5 shots covers the dr of the scene. the human eye can see a scene with a dynamic range of more than 20 stops. even a 5 shot bracket will not even touch this kind of range. again this is why you have to spotmeter thje scene so you know what you have to cover in dr before you begin shooting.

the best example i can give is a sunset with the sun just below the horizon by about 10-20minutes. a 5 shot bracket with not begin to cover the dr of that sunset scene. you are probably talking somewhere in the area of 15shots to bracket maybe more to cover the dr.

my hdr how to was a written set of instructions to newbies who wish to know the way to make an hdr image without any shortcuts. i know it works because that is how i make them. i also know that if a newbie follows the info they would get a good hdr final image. i also tried to take all the variables into account when writing it. the scene may or may not require all the info to make it work, but i wanted the reader to have it all if needed.

 

[GaryDeM]

"With PSPX2, can you adjust each raw for eg. WB which setting is retained when you then send each raw into the merge?"

no you cannot adjust any raw shots individually. if you did then the merge would show the same areas with different views. if you had 5 shots in raw and used acr to get better shadow areas in each, which one would the software use to make the final image?

getting more dr is why we are doing hdr there is no need to hand fix each shot. if done right with enough bracketed shots the hdr will take care of the problem.

as for wb you set awb as the seting or go with a preset setting for all. the problem comes with the merge, which different color image is the hdr software going to use?

for myself, i shoot the hdr shots in jpeg. this means that all ARE the same because the camera's settings are the same.

this is, of course, why the dslr's jpeg settings should be setup before use to give the best and most accurate image possible. you do not have to be concerned about the dr of jpegs because the hdr process will take of that.

 

[PicOne]

as for wb you set awb as the seting or go with a preset setting for
all. the problem comes with the merge, which different color image is
the hdr software going to use?

So, what do you do if you decide that the WB chosen by AWB, or preselected setting, is not ideal? IOW, if you adjust WB on a raw shot in PSP, and then close or resave the raw file, isn't that WB setting retained? I wasn't really considering that you'd apply different WB settings to each Raw file that will be in the merge, but that you would set all of the shots WB to the same setting (but perhaps different than "as shot").

Likewise, with any adjustment available in a Raw converter (eg. LR) -- why couldn't you apply different eg. Saturation, Sharpness, noise reduction, Fill light, tint, hue to the Raws prior to them entering the HDR merge?.. if need be you do the same settings to all of the Raws to keep consistent. In the end, whether 1 raw file by itself, or 5 merged raw files, the software is applying its own "default" settings anyway to each -- why not consider moving off of these defaults?
--
'Everything in photography boils down to what's sharp and what's fuzzy.'
-Gaylord Herron