Finished shot. Nikon D800E, Zeiss ZF.2 2.8/21 Distagon – composite of three images
This post is for those of you who are curious about how much work exactly goes on during an architectural/ interior shoot. (And yes, the company whose boutique is featured here is one of my clients, as is their parent company.) It’s not quite as simple as it looks. First, let’s look at the original, unedited shot again – this is what you get out of camera with the raw file converted straight to a JPEG, all ACR adjustments zeroed out. This is about as good as you’re going to get straight out of the camera. Light was available ambient, and I used the excellent Gitzo GT5562LTS 6x Systematic carbon tripod for support.
The same shot, straight out of camera, NEF>JPEG via ACR with all adjustments set at neutral.
Can’t see any issues with this image? Let me point them out to you.
- There’s a bit of sombrero distortion in the horizontal lines; this is a lens issue. Minor, but obvious when you’ve got this many parallel lines.
- The horizontals run to the right slightly; it’s almost impossible to get this perfect in camera, even using the built-in level guide – as with the in-focus indicator, there’s a bit of ‘dead space’ around the null position where the camera might report perfect levelling even though things are a degree or so off.
- The highlights are gone. All gone.
- The shadows are also all gone…oh dear, this is turning into a tonal nightmare: very dark wood on the storefront, extremely brightly-lit illuminated displays…
- Very distracting reflections from the store and the boutique across the hallway – we couldn’t get them to turn off their storefront lights (obviously) – so there was no choice but to retouch them out.
Now you know what I’m up against, let me explain how the issues were solved. Distortion is taken care of using ACR’s built-in profile for the lens; note how everything is straight again in the finished image below. Slight issues with levelling are taken care of using Photoshop’s distort tool – select all, then Edit>Transform>Distort. Use guides to help you line up your horizontals and verticals, and then pull the corners a little bit to make everything line up.
Dynamic range is a bit more of a challenge. Although the rough image above has had zero highlight and shadow recovery done and perhaps represents about 8 bits of tonal information, I wouldn’t want to push this image too far in post. The extreme highlights will posterize slightly and the shadows will get noisy, though I might be able to recover 13-14 stops this way. The image above was actually exposed for the midtones. The shadows and highlights had separate images three stops apart. I overlaid one on the other, and erased out the bits I didn’t want – so that would be putting the darkest image on top, then erasing around the highlights to reveal the images below. This is the correct way to do HDR – and perhaps the subject of a future article. The tonal transition between layers was made smooth by feathering the opacity and hardness of the eraser brush, as well as using curves on the individual layers afterwards.
Two more steps – dodge and burn of the flattened image to give it local pop; and the hardest bit: retouching out the reflections of the opposite boutique. Note that the reflections overlaid some very complex textures and structures in the image; it definitely wasn’t a simple clone or healing brush job. (Originally, the images were done in the morning; the boutique opposite proved to be extremely bright, so I came back at 10pm once most – but sadly not all – of the lights in the neighbouring boutiques had been turned off. ) The solution? Get creative with replacement textures. The left hand window is actually an inverted clone of the right hand window; everything was nearly perfectly square geometrically, so it wasn’t too much of an issue. Small differences in alignment were sorted out using the Distort tool again, matching up the corners. It also helped that the interior of the store itself was mostly symmetric, of course. Some strategic erasing around the left-unique features and some healing brush later, and voila – finished.
As much as I like to get it right straight out of the camera, there are some times when it simply just isn’t possible. MT
Finished shot again.
____________
If you enjoyed this post, please consider supporting the site via Paypal (mingthein2@gmail.com); Ming Thein’s Email School of Photography – learn exactly what you want to learn, when you want to learn it or learn how to achieve a similar look with our Photoshop workflow DVDs. You can also get your gear from Amazon.com via this referral link. 
Don’t forget to like us on Facebook and join the Flickr group!
Images and content copyright Ming Thein | mingthein.com 2012 onwards. All rights reserved
Ming Thein | Photographer 
Never knew so much went into one shot!
Nor do a lot of clients – hence the post!
Ming, you write what I consider the most thoughtful and perhaps incisive photoblog out there. It is always a pleasure to check in! I particularly enjoyed your 10×10 article — that was a lot of work to put into making better photographers out of your readers. Mostly I do a lot of professional architectural work, though I have spent time as a war photog and done jobs in the product, fashion, and portrait fields. A brief note on your reflections issue (though I liked your solution and have done the same), if you had two light stands and A-Clamps handy for this job and a big black cloth (or actually a roll of black plastic — think garbage bag type plastic that in the states you can pick up at a home depot in 30ft x 12ft rolls), you can position the plastic or cloth behind the tripod raised between two stands to eliminate bad reflections on storefronts. For example, I’ve had to do this when shooting restaurants in casinos. I know this might not have been practical depending on the street front, but on the off chance you haven’t used the technique, it can be handy for more than a couple of scenarios.
Thanks Tyson. That’s a great idea – I didn’t think of it actually. No idea how the other store would have reacted though…
nice picture.
However, there where on the right side, the shadows on the dark part of the ceiling (there where the spotlights are located) line up with the curtains, their mirrored copy on the left side doesn’t match up with the surrounding parts; there’s nothing that might explain those shadows.
Now that I’ve noticed this, it draws my eye every time I look at the picture…
Good spot! I did notice this but there was no way of cleaning it up to mimic the reflection of the actual objects. I don’t think 99.9% of viewers will see it, however.
This is probably the oldest HDR technique (dating back even into the early years of analog photography), however this is NOT the only correct way. It only works well if the parts of your image that need to be levelled are seperated from each other, like with windows. It is perfectly fine and actually technically advanced to let an application create a real HDR image (that is NOT the tonemapped output) because this way, the tonal spectrum is extended througout the whole image and any subsequent exposure compression is applied to a greater area or even the whole image rather than just a local area. In short, this gives you much more control over the way the exposure is compressed.
YES, there are tons of badly tonemapped “HDR images”. But that’s the fault of the user, not of the merging technology. (Just as your overlay technique can lead to bad results as well if the user doesn’t master it.) Frankly, this HDR snobism is just as awful as those badly tonemapped HDR images. I love reading your articles, but please keep this upcoming article technical and do not bash on more modern HDR merging technologies when you are clearly biased by bad examples.
Yes and no…this technique can also be used in combination with feathering, different blend modes and soft-edged masks to create the same effect as an application-generated HDR, but with more specific local control – even if you don’t have hard edges. In fact, I had to use it here in places to preserve the highlights in the wood panelling.
My definition of ‘HDR done wrong’ refers to badly tonemapped images, not the technique per se. I just don’t use the automated software because I don’t like the lack of control over the output tone.
Great work Ming,” local pop ” any usage of pluggins like Lucie recommended. most of your work just pop. Thanks
Nope, I don’t use plugins. All Photoshop only – I’ve got a DVD covering my workflow available here 🙂
Thank you for this interesting post. Learnt allot again. not in the least to better look at photo’s because I only noticed the difference in display set up after you pointed out that it was a inverted clone 🙂
Regarding the reflections, could have helped a polarizer?
No, because most of them were exactly at 90deg – i.e. behind the camera – not at an angle, so it wouldn’t have made that much difference.
You’re right Ming, this techniques is best for natural HDR.
I bought many kind of hdr software, but never liked it.
Photoshop layers are the best way to do. Did you tried blend if too.
Nope, just careful use of layers with strategic erasing and different opacity.
Hello Ming; I am going to echo Daniel here:
““This is the correct way to do HDR “
Definitely interested in this, too. Cheers, KL
I think it’ll have to be in the next set of photoshop ‘advanced’ techniques…maybe time to make another DVD (after the Photokina madness ends, at any rate).
I’m going to quote just to emphasize
“This is the correct way to do HDR “
Thank you!