Noob Question

Hi Everyone ..

I read a very interesting post in another thread (attached) that have piqued my interest in HDR (again), and have a couple of noob questions; I hope I can find some answers here.

Q1. If a RAW image has every bit of information that can fall on the sensor at the time of exposure, why do we have to have more than one image? Can't you just (via software) subtract/add 1 stop, copy the image, and merge 'em? I suppose that's what a single image HDR is, but then, why do we need more than one?

Q2. When I shoot HDR, for some reason I have always gone -1~0~+1 but everywhere I have been reading recently, the examples all say -2~0~+2 .. So have I not been getting the most out of my efforts by setting the range too low? Should I be using -2~0~+2?

Q3. Is there a time you'd use -1~0~+1?

Thisis great advice ..

imjason wrote in post #14586759
im more interested in why youre shooting at f22 instead of say f16 on an ultrawide? whether your subject is 10ft away or 1000ft away, youre not gaining much in DOF (only 3-4inches at 10mm). Instead youre probably softening the image and slowing down the shutterspeed.

anyways, ive done 9 shot HDRs, using the custom settings shouldnt be a problem. you can also just stay in M mode and just roll your shutterspeed or EC in AV without the need to switch to your custom modes. the only thing i can think of is that you might shake the camera or turn your camera slightly left or right when you twist the mode dial. since youre capturing all these images in quick succession, the slower shutter speed from using f22 might cause slightly blurred images. Ive done HDR with filters such as a CPL. they really make things like clouds pop even more than usual.

if you want more range, you an go: -4, -3, -2, -1, 0, +1, +2, +3, +4.
how i do it:
1. AEB set at +2/-2
2. find the 0 EV.
3. set shutter speed to -2ev, take first 3 shots.
4. set shutter speed to -1ev, take second 3 shots
5. set shutter speed to +1ev, take third 3 shots
6. set shutter speed to +2ev, take last 3 shots.

You can definitely use the RAW file to get +/-1 EV images to run through your HDR program. You can even run a single image through. The reason you bracket is because you get more detail at the longer exposures by shooting rather than taking it out of the RAW file. The RAW file doesn't have 100% of the detail for all exposures within the range. You can pull some detail out, but not all. The purpose of the method is to even out the exposure across high dynamic range situations, so I don't see really why someone would be shooting +/-1 when you should be using it in situations where the difference between the lights and darks is more like 4-8 stops. I usually do +/- 2 EV with the default camera software. I started using Magic Lantern recently, which allows me to shoot up to 9 (I think) bracketed shots. For this I might use +/- 1 EV if I am shooting 5-7 shots depending on the range of the scene. It really all depends on the range of the scene you are shooting, but you will definitely get better results using bracketed shots from the camera rather than creating the bracketed shots from a single RAW file.

yo!

a single raw file only includes information for the single exposure. for example, if your highlights are blown out, no matter how fancy your PP software may be, it will be blown as that is how the camera recorded the scene. In order to create an HDR photo without the blown out area, one takes another shot with a faster shutter speed. this could be -1ev, -2ev or even -4ev. the point with bracketing 3 shots is to allow you to create a HDR image that will cover greater dynamic range greater than what one shot can produce, hence High Dynamic Range.

imjason wrote in post #14607571
a single raw file only includes information for the single exposure. for example, if your highlights are blown out, no matter how fancy your PP software may be, it will be blown as that is how the camera recorded the scene.

Ok, I am definitely picking up what you just laid down; thanks guys

Your bracket must cover the dynamic range of the scene you are trying to capture. Using a prescribed approach such as "-1/0/+1" is irrelevant unless it is appropriate for the scene. The SCENE you are shooting dictates the bracket you should capture. The interval between exposures in your bracket dictates how accurate your merge, luminance and color rendering will be.

If your scene requires -4 / 0 / +4 you can try to cover that in 3 shots, but most HDR merge strategies will collapse and produce awful results with such large spacing between exposures. In that case try, -4, -2, 0, +2, +4 and see if that gives you better results. Etc.

When you use -1, 0, +1 you typically have a scene that is almost well exposed in a single exposure, but needs extra highlight protection. Nothing wrong with that.

See:

http://www.hdrlabs.com​/tutorials/index.html

for a nice compendium of tips.

kirk

Because of my relatively untrained eye, what should I do to get an accurate assessment of the range that's present in my compo? Is it as simple as in camera metering of the brightest and the darkest spots in the image and summing them?

Set your camera to use spot metering mode. Set the camera shooting mode to Manual. Zoom in, if you are using a zoom lens. Spot meter your brightest highlights that contain detail (i.e., not specular reflections off of chrome). Spot meter your deepest shadows that contain detail. Spot meter something in the middle that you want to be considered middle gray. These values will give you the shutter speeds you need to consider to bracket.

In case you are a little unfamiliar - when you spot meter in manual mode, you will be using a fixed aperture and rolling the shutter speed until your exposure indicator is at "0". This means that whatever you are pointing the spot meter at is now exposed to middle gray (i.e., "proper exposure"). Note the shutter speed that gives you this proper exposure. When you shoot HDR sequences, you want to hold ISO and aperture fixed and vary shutter speed - thus, think of exposure brackets in terms of shutter speed. 1 EV increments of shutter speed will be something like:

1s, 0.5s, 1/4, 1/8, 1/15, 1/30, 1/60, 1/125, 1/250, 1/500, 1/1000, 1/2000, 1/4000

So, if you fix your ISO at 100 and your aperture at f/8, you might get a scene that meters like this:

highlights: 1/2000
shadows: 1/8

maybe your desired mid tone is at 1/125 or so.

Then, you need coverage over that range - in 1 EV increments, this would be:

1/8, 1/15, 1/30, 1/60, 1/125, 1/250, 1/500, 1/1000, 1/2000

If you know how much coverage you get form a single shot (say 8-9 usable EV steps) then all you may need is -2, 0, +2 to cover everything - for example:

"0" exposure - 1/125
"-2" exposure - 1/30 - this will extend the shadow end of the range (here I am using "-2" to mean the EV step I increment the camera's shutter speed)
"+2" exposure - 1/500 - this will extend the highlight range

If my camera captures 8 EV of DR in a well exposed 0 EV image, then the 0 EV image here, taken at 1/125, will cover about 1/15 to 1/1000 or so. The + and - images will extend the range 2 EV above and below, covering the extra 1/8 and /1000 with data to spare. Some folks like to capture redundant images (e.g., -2, -1, 0, +1, +2). Same range, finer increments in the data.

It depends on your camera and your workflow - you may have shadow DR that becomes unusable very quickly due to noise or other factors, etc. The super data intensive route would be simply to shoot an exposure at each shutter speed. Then you can go back to your computer and decide later which exposures you want to use, instead of regretting that your minimalist approach hosed you a little bit on the shadow or highlight end of the range.

have fun!

kirk

Kirk, thanks for the lesson; I appreciate it greatly; but as a noob, I have more questions

1. I've tried, but am not able to find the DR of my 5D2, 7D, or 5D; can you please point me in the right direction?

2. You said zoom and meter; this is more accurate than just metering at the wider angle I will be shooting at?

3. Theoretically, I could just take all 9 of those shots and stuff them into the software .. but is there a point of diminishing returns? You said, "you may have shadow DR that becomes unusable very quickly due to noise or other factors, etc." Could you get undesirable results from redundant or noisy data, or is the currently available software (Photomatix etc.) capable of understanding and discarding redundant/noisy data? I know it allows noise reduction, but it is better not to have noise in the first place, right?

4. Lately I have been thinking about getting a EFS 10-22 and using my 7D for landscapes (just because I like the newer menus and the ergonomics of it). It does get noisy at 800+ but at ISO 100, it should be apples and apples with my 5D2, right?

I will try this out as soon as we have a good day come up but for the past couple of weeks we've had oppressive heat in the central KY region and its supposed to go on for a while.

Thanks again your your time and the valuable info. I printed out your response and am absorbing it; someday I hope to be able to do this in my head. When does the learning end

Freddy ..

P.S. Now I understand why I have all the over exposed stuff in this otherwise pretty good image; my default has always been -1/0/+1 but I guess the highlights were still too bright.

IMAGE LINK: http://www.flickr.com/​photos/kykhans/7098842​295/

IMAGE LINK: http://www.flickr.com/​photos/kykhans/6921952​880/

alphamalex wrote in post #14612757
Kirk, thanks for the lesson; I appreciate it greatly; but as a noob, I have more questions

1. I've tried, but am not able to find the DR of my 5D2, 7D, or 5D; can you please point me in the right direction?

DR = dynamic range. Basically the difference in exposure between the light and dark areas of the scene you intend to shoot. It's not a camera setting. Evaluating the DR is done with the spot metering process Kirk laid out.

alphamalex wrote in post #14612757
2. You said zoom and meter; this is more accurate than just metering at the wider angle I will be shooting at?

Usually yes. Remember that the spot meter evaluates the exposure at the center of the viewer. If the lightest point is small in your overall scene, you could get a darker exposure than you really want when spotting the light area. If you zoom and fill the center with the light area, you will get a truer evaluation. But if you are just evaluating the sky in general as the light area and there is no significant change in the dynamic range in the sky across your scene (i.e. you have the sun in or close to your frame), then zooming probably won't make a difference. It depends on what you are shooting.

alphamalex wrote in post #14612757
3. Theoretically, I could just take all 9 of those shots and stuff them into the software .. but is there a point of diminishing returns? You said, "you may have shadow DR that becomes unusable very quickly due to noise or other factors, etc." Could you get undesirable results from redundant or noisy data, or is the currently available software (Photomatix etc.) capable of understanding and discarding redundant/noisy data? I know it allows noise reduction, but it is better not to have noise in the first place, right?

The HDR process creates noise, so you don't want any inherent noise in your photo. HDR will make it worse. A RAW file has detail at EVs other than the one you shot it at, but that detail is not the same as if you had shot the photo at the lower exposure. If you shoot a photo at -2EV and create an image at -2 EV from the RAW file, the -2EV photo will have more detail and less noise.

alphamalex wrote in post #14612757
4. Lately I have been thinking about getting a EFS 10-22 and using my 7D for landscapes (just because I like the newer menus and the ergonomics of it). It does get noisy at 800+ but at ISO 100, it should be apples and apples with my 5D2, right?

The 5D still takes a better picture since it's full frame, so the sensor captures the light better. The noise you are worried about isn't necessarily cause by the ISO setting, but in the process of trying to bring back detail from an underexposed area of the picture. If you take an underexposed shot and try to bring back the detail in PS/LR, you are often going to have noise at whatever ISO you use. The point is that you want to take one of the pictures in your bracket for your HDR photo at the correct exposure for the details in the dark area of the scene that you are shooting. So when you merge the photos, the software will look for the best exposure for every area in the scene and you will have it in one of your photos. Hope that makes sense

alphamalex wrote in post #14612757
P.S. Now I understand why I have all the over exposed stuff in this otherwise pretty good image; my default has always been -1/0/+1 but I guess the highlights were still too bright.

I like the images, but the top one definitely has some blown out areas. If you are just taking a single photo, learn to check the histogram for "blinkies". Hit the play button to view the photo you just shot, then info to see the brightness histogram. If there are areas of your photo that are blown out, you will see those areas blinking. You will also see some of the histogram against the right side of the graph. If you want to optimize your shot, you should be "shooting to the right", that is creating a picture where most of the area of the histogram is to the right of the center of the graph and with no "blinkies". That way you don't have overexposed areas, but you preserve as much detail as possible in the dark areas. This is just a general tip for landscape, not something to do with HDR.

alphamalex wrote in post #14611706
... what should I do to get an accurate assessment of the range that's present in my compo?

With aperture and ISO constant, I monitor the live view histogram (exposure simulation enabled) as I change shutter speeds. Start with a shutter speed that covers the shadows (left side of histo) and decrease the exposure a stop or two at a time 'til the highlights are covered. Done. This, of course, requires a sturdy tripod.

The dynamic range of all three of your cameras at lower ISOs is about 11EV (this is an evaluation of the sensor, not the total optical system. See:

http://www.dxomark.com …il3)/176|0/(bra​nd3)/Canon

A safe assumption is that for ISO 100-200, you can probably capture about 10EV in a single exposure. See the attached chart for a sense of how these three cameras DR performance changes as a function of ISO. You can find a ton of data at the DXOMark website:

www.dxomark.com

and nerd out to your heart's content!

As far as spot metering goes, the spot meter of your camera is typically looking at the central 1-3 degrees of the field of your camera. If you can zoom, then you can get a smaller "spot" relative to the scene and make a more accurate measurement that is not including tonal values outside of those you want to meter. Here is a nice short blurb about spot metering that I just came across that has some interesting little tidbits:

http://laurencekim.com …for-portrait-photography/

and there is always the Sekonic website:

http://www.sekonic.com …/IncidentvsRefl​ected.aspx

As far as the merge - more is not always better. But, shooting a 1EV sequence across the entire range of shutter speeds does not mean you have to use all of the images, it simply affords you more data to chose from. The merge method will also affect the data's being prone to noise and color shifts, so know the merge method you are using and feed it the food it needs to get the job done.

I have not shot with a 7D so I do not know how it compares to a 5DII. I have a 5D and a 5DII. I have made HDR image data with 15mm ff fisheye all the way to a 200mm. If you are shooting landscape, there is probably sufficient light to shoot at lower ISO, so, regardless of how the 7D and 5DII compare, the results for the 7D at lower ISOs should be fine.

The top image you posted is just destroyed by the blown sky, not just because it has blown highlights, but because the bokeh has blurred and spread them into a glaring distracting mess. The remainder of the image is nice. THe bottom image is real nice, although there is some haloing at the horizon, maybe due to sharpening radius being a little too high.

The HDRLabs site is a great resource. I would urge you to download and use the FREE Picturenaut application, as the whole workflow, especially in the merging stage, is exposed and not hidden - you can learn a lot even if you end up using other applications.

kirk

Dang, this is a lot to digest; thanks everyone .. now to see if I can put this to good use. I gotta take sweet time and take the tripod too; I'm always in a rush.

BTW, I should focus at one point and then leave it the heck alone, right? I do have remotes for all my bodies, so I'll just set lens to manual and use remotes or 2 sec timer without.

Now to learn about hyperfocal distance l so I can choose the right place to focus. In this long exposure HDR, I somehow managed to get a lot in focus un-intentionally

IMAGE LINK: http://www.flickr.com/​photos/kykhans/6922204​010/

I think the general rule is to focus on something about 1/3 of the way between the nearest and the farthest point you want to be sharp. BTW, you have a lot of nice gear for a "noob".

shhh ...

I think she likes this hobby better than my last one which was amplifiers and speakers; this way she gets all these "oooh, aaah, you look so cute" compliments from her little facebook friends.

Also, once every room in the house had tunes, I had to find another hobby

I'd like to take this hobby to the next level, but that requires time and effort; with three kids in elementary school, a full time job, and gigs on the side (software design/development), I can only give 20 minutes a day to photography, which is not fair to me, or the hobby ... Heck, I just want to earn enough to pay for the gear. Maybe a coffee table book, or something is in the future, who knows

On a serious note, I remembered about the one third between the near and far points yardstick; thanks for the reminder. I will read up on it.

Just curious- how do you measure a third of infinity?



kirk