Hey Pro's I need help regarding HDR Photography using Single Image

[demyusinvahd]

Copyright - Pipoyjohn PhotoArt

Like this one. :|


Help me Sir Godbless.


Can anyone give me a workflow?

[Manfred M]

First of all it is, by definition (almost) impossible to create a HDR image out of a single image. I think most of the people on this site are not too impressed by overprocessed images like the one shown here. To get what you want you will have to acquire some HDR software and run your images through it. Photomatix is probably the most popular package out there right now. I personally prefer the Oloneo Photoengine and NIX software's Hdr FX has gotten good reviews too.

To get a true HDRI image you need at least two different images; one over exposed and one underexposed that can be combined to show a broader tonal range. You can sort of do this from a single image if you have a RAW file. You need to prepare three tiff images; one with the lower range (darks), one with the higher ranges (lights) and one with the mid-ranges. A jpg will not work because you only have 8-bit data. The 12 or 14 bit RAW files let you get at data that is thrown away during the conversion to jpg. Combining them with HDRI software lets you access the entire dynamic range that your camera's sensor can produce and will allow you to at least partially create a HDR image / effects. These images do have constrained dynamic range when compared to the traditional multi-shot process.

The problem with the image that you show is that it is not a particularly good image to start with and applying HDRI does make it look different. Different does not make it look better, and in my opinion, is not something that works here. People seem to love running perfectly ordinary images through the "grunge" setting and the internet is flooded with images similar to this one.

I have played around with HDRI and have yet to get a result that I am really happy with. In all cases I only resorted to the technique when I faced shooting conditions that my camera's sensor could not handle and the results were pretty awful (blown highlights and lack of shadow detai). No use of the "grunge" setting either. These examples are all either sunrise or sunset shots and are true HDRI.

[pnodrog]

I expose one for highlights and another for shadow detail then I merge the two using a layer mask. It is more time consuming than using HDR software but I like the absolute control and it avoids it looking like many software merged HDR images. Actually when I started doing it with shots scanned from Fuji velvia film the software was not readily available. If it looks like HDR as far as I am concerned it has been over done. Subtly is the key - solving HDR scenes is a problem the photographer needs to overcome - if the viewer is unaware of the techniques to produce the image the photographer has succeeded. The tree shot immediately above has succeeded.

Read the tutorial on Digital camera vs the eye. When humans (ignore photographers) view an HDR scene their eyes and mind do the all the adjustments they need. We just need to try duplicate that processing using a camera and processing software.

The trick in photography is to produce what they thought they saw.

[Donald]

Like has been said above, an image made from a single frame cannot, by definition, be one that applies HDR principles. HDR results from a scene containing too great a dynamic range to be captured in one file and requires the capturing of two or more photographs that are then blended together, sometimes using software specifically designed to do the job. If you've captured all the data in one file, then you cannot apply HDR principles and techniques.

You can make different versions of that one file and then blend them in order to create a particular effect, usually over tone-mapped. But you do not have an image that is the result of applying HDR principles.

[FrankMi]

If you want to learn how to use HDR techniques, it would be better to start with learning the basics, which includes identification of what kinds of images can make effective use of the Tonemapping process. Here is a free tutorial that can help get you started.

http://www.stuckincustoms.com/hdr-tutorial/

[ajohnw]

I'm curious as I thought HDR and tone mapping were entirely different processes?

On the shot it's easy to get so interested in the amount of sky in the shot and finish up with buildings that are falling over. Can be corrected but it's still worth bearing in mind.

John

[Manfred M]

I'm curious as I thought HDR and tone mapping were entirely different processes?

On the shot it's easy to get so interested in the amount of sky in the shot and finish up with buildings that are falling over. Can be corrected but it's still worth bearing in mind.

John

HDRI uses tone mapping to create the final image.

If you think about it, it does make sense. You are taking a number of images that exceed the dynamic range of the output device and try to display them. The only think that can be done is to take the extreme values (and everything in between) and bring them down to the level where they can be displayed without blown highlights and blocked shadow detail; i.e. tone mapping.

[FrankMi]

I'm curious as I thought HDR and tone mapping were entirely different processes?

In one sense, they are different. HDR (High Dynamic Range) describes the condition of the scene; it has a high dynamic (contrast) range. Tonemapping is a processing technique that can be used to compress that range so that the details in the otherwise blacked-out shadows and blown-out highlights can be seen in a single image. You usually apply tonemapping techniques to multiple duplicate images with bracketed exposures.

When tonemapping is inappropriately applied, particularly to a single image, the results can range from ineffectual to downright bizarre.

When tonemapping is appropriately applied, the results can be quite pleasing but in order to get those results, significant post processing is often required after the tonemapping process.

[ajohnw]

Thanks. I have only played with HDR on one package - Fotoxx. It presents the images as a light pixels to dark pixels graph for up to 9 images. With a 2 shot HDR the highlight one slopes one way and the brightened image the the other way. The 2 can then be mixed to varying degrees by altering the shape of the curves. I just did one for fun. 2 shots from a raw. Not uploaded on here. I feel I need to be proud of them to do that.. I've deliberately allowed the brightened exposure to move the water towards over exposure. It relates to how the image would look with a better exposure of the duck.

http://www.flickr.com/photos/48994619@N08/7923481540/

I then used tone mapping but to me that is another graph representing differing contrast levels in the shot from low to high contrast allowing them to be changed accordingly. Raise the low contrast end and low contrast detail will have it's contrast increased and so on. I think I am getting the hang of this but had to denoise a bit to reduce processing noise I had introduced into the ducks bill. In some ways much of this is down to using a Pen. I generally only have to use curves to bring out dark areas a bit when I use my canon or compact. The shot also needed defringing - thanks to the Pen's sensor I suspect.

I wonder if the difference is down to Gimp speak as against Adobe? Gimp will do the same things as Fotoxx but can be a lot more complex to use. Maybe this style of HDR is a different method or I have misunderstood what is going on.

John

[FrankMi]

Hi John,

There may very well be a difference in the way Photomatix processes Tone-mapped images. Perhaps a before and after image will help?

This first High Dynamic Range image is the SOOC normal exposure of a three image bracketed exposure set:



And this is the same image Tone-mapped in Photomatix from the three bracketed exposure HDR images plus some clean-up in Photoshop:



The first is an HDR 'image' and the second is the result of the Tone-mapping process.

Does this help at all?

[ajohnw]

Yes it looks like an entirely different process. Fotoxx and one of the Gimp plugins uses the exposures as layers and then offers the ability to mix them according to brightness across each shots range - layers and transparency based on brightness. I worked with these

Camera jpg - what a "normal" raw development would give trying to get water and the duck correct. Not used at all,

http://www.flickr.com/photos/4899461...7631399852866/

The raw development for the water

http://www.flickr.com/photos/4899461...7631399852866/

And then the raw development for the duck

http://www.flickr.com/photos/4899461...7631399852866/

These were intelligent ducks that associated people by seats with food which meant the lighting was such that normally I wouldn't bother. It's interesting to see what came out of one raw file fairly easily and it could be improved further by processing the 2 shots used before combining them. I didn't keep the image formed when they were combined. Post processing from memory was tone mapping to increase low to middle contrast, mostly low, Weiner sharpening, weak textural contrast, slight increase in saturation and defringing the final sized shot. I suspect that should have been done on the duck exposure, Lastly a little noise reduction as I had over processed slightly. Forgot to keep a close eye on the duck's bill. I could have smoothed that locally but was only really trying the software used out.

I rather like Fotoxx. One of it's interesting aspect is it's plugins. These are any other applications making it easy to use specialised packages when needed.

The Pen used - a lot easier to carry around than my canon gear. I probably walked 4 or 5 miles on that day. One camera, 2 lenses covering the equivalent of 28 to 400mm in 35mm terms. The duck shot is a crop so more equivalent to 800mm really. A bit more thought and Olympus would have a great camera rather than one with a number of niggles. Some serious and some just a pain.

John

[FrankMi]

One critical aspect that makes Tone-mapping difficult, if not impossible with a typical multiple bracketed exposures of something like a duck is that everything is moving from image to image. It would be VERY difficult to align the images as I did with the one room colonial home.

If you take a single HDR image and make lighter and darker copies then combine them, you are still limited by the lost detail in both the saturated shadows and blown highlights. The result is usually very grainy owing to the lack of detail in the shadows and highlights.

If the image was able to capture the entire contrast range without saturated shadows or blown highlights, then it isn't really HDR and can be processed without having to revert to Tone-mapping techniques.

In the end, you can use any technique you like on any image. If you are happy with the result then it is, for you, successful.

To put it another way, a carpenter doesn’t buy a drill because he wants a drill. He buys a drill because he wants a hole! In the end, the result is all about ‘the hole’, not the drill and only the carpenter cares which drill and drilling technique was used to create the hole.

[jcuknz]

When I first heard about HDR ... that is HDR done with moderation and not the garish examples ....I wondered what the fuss was about because I've known and done similar with single shots for ages.
The approach I normally use is to use an adjustment layer with different degrees of grey for different areas of the photo ...
One sets the AL for the most adjustment needed and then modify other ares back with different amounts of grey to black for areas needing no adjustment.

[ajohnw]

I did this one the hard way effectively dodging it for brightness,contrast and saturation etc.

http://backup.cambridgeincolour.com/...43597/original

The 2 shots from raw was much quicker and really this one would be a tougher test. as I suspect it would need extra exposure for the head. Probably more worth the effort as well.

What ever method is used the end result is the same - a bigger colour space reduced to sRGB largely because of light levels. To me that is HDR how ever it's done. A straight reduction to PC colour space will result in areas being too light or too dark so multiple "exposures" are needed. HDR is a technique.

John

[demyusinvahd]

Thanks everyone for your response !

[Colin Southern]

Like has been said above, an image made from a single frame cannot, by definition, be one that applies HDR principles. HDR results from a scene containing too great a dynamic range to be captured in one file and requires the capturing of two or more photographs that are then blended together, sometimes using software specifically designed to do the job. If you've captured all the data in one file, then you cannot apply HDR principles and techniques.

You can make different versions of that one file and then blend them in order to create a particular effect, usually over tone-mapped. But you do not have an image that is the result of applying HDR principles.

Hi Donald,

The one exception to the rule (hell, may as well just about call it MY rule!) is when one uses GND filters to optically compress the dynamic range into something the sensor can handle in a single exposure.

[ajohnw]

Must admit some people refer to an hdr created from 2 images from a single raw file a fake hdr.

John

[Manfred M]

Must admit some people refer to an hdr created from 2 images from a single raw file a fake hdr.

John

I agree with your statement, as I've tried it. Essentially what you are doing is emulating a RAW to jpeg conversion, but taking a different tool (the HDRI software) and different "views" of the same RAW image. No, it is not true HDRI as you are still sensor limited, but neither is it purely a single image tone-mapped image. It's really something in the middle.

[Colin Southern]

I agree with your statement, as I've tried it. Essentially what you are doing is emulating a RAW to jpeg conversion, but taking a different tool (the HDRI software) and different "views" of the same RAW image. No, it is not true HDRI as you are still sensor limited, but neither is it purely a single image tone-mapped image. It's really something in the middle.

Its really just a processing technique. We all know that it still doesn't contain any more info than just one of the RAW files - but - it CAN make it easier to push/pull the levels to where you want them when compared to trying to do it from just a single file. That's my experience anyway

[Glenn NK]

Its really just a processing technique. We all know that it still doesn't contain any more info than just one of the RAW files - but - it CAN make it easier to push/pull the levels to where you want them when compared to trying to do it from just a single file. That's my experience anyway

Exactly.

I've gone back over older files with the latest version of Lightroom (4.1) and my impression is that I'm able to pull details out of the shadows and highlights that I could not do with previous versions. This of course would seem to apply to CS6 (which I don't have).

If one can get a reasonable result by combining two version of the same image (fake HDR), then it begs the question as to why not work with the single original image?

Comments anyone?

Glenn

PS: It's possible that with the future (probable) advancements of sensor technology in conjunction with advancements in processing ability, HDR won't be necessary at all.