Wednesday, May 12, 2010

My HDR Work-flow

Welcome to my explanation of how I process pictures.  I hope what I've explained here will be easy to understand and I hope you might be able to use some, if not all of what I do here.  If you have time, please take a read.  If you find something difficult to understand or have something you'd like to add or would like me to add, please don't be afraid to comment.  I would really appreciate it.  I may tweak this from time to time and when I do, I will certainly let you know.  Also, if you think screen-shots would help, let me know and I may add some in later.  I just didn't feel they were necessary.  This is really a hybrid work-flow and tutorial, sans screen-shots.

This is a synopsis of how I GENERALLY process most of my pictures.  There will be times when this processing does not work and a different approach is needed, but I find this really works for me most of the time.  Before I get started, I must point out a couple of quick caveats to this. First, everyone's camera is different and what comes out of it may a look a little different than what comes out of mine.  I haven't processed anyone else's shots using my formula so I can't guarantee you will get the same look.  However, I think you will get something quite close.  The other caveat is that the processing I use here will strip off the exif data from your final product.  It actually happens during the "merge to hdr" process in Photoshop CS4.  I'm hoping the new version, CS5, will mitigate that problem since I hear it has a more robust HDR processing engine in it.  However, I don't know for sure if it is fixed or not, but I hope to find out someday when I upgrade. Lastly, I usually take 5 bracketed raws ranging from -2 to +2 EVs in 1 stop increments.  The minimum of three will work fine as well, but I prefer 5. With that said, on to the workflow...

Tools I use are as follows:

Photoshop CS4
Photomatix Pro 3.2.6
Nik Color Efex Pro 3 Complete
Nik Sharpener Pro 3
Imagenomic's Noiseware Pro

1. Merge photos in Photoshop CS4
    A. Open Adobe Bridge and select all brackets to use.  Open all in Adobe Camera Raw.  Make any adjustments you see necessary.  I usually click on auto white balance.  That's about the extent of it for me.  Then click done.
   B.  With all of the brackets selected, go to Tools | Photoshop | Merge to HDR...  I feel the alignment engine in Photoshop works better overall than it does in Photomatix.  This is a trick I learned in Scott Kelby's CS4 book.  Once the merge is complete, save it as an .hdr file.  This is the process that strips the exif data that I mentioned in the caveats earlier.  It sucks, but I live with it...

2. In Photomatix, I open the .hdr file I just created with Photoshop and tonemap it with Details Enhancer. The settings I use are as follows (but sometimes I will vary them slightly, but these are usually what I use):

Strength: 95
Color Saturation: 76
Luminosity: 9.1
Microcontrast: 0.0
Smoothing: put a check mark in Light Mode and select better look between High and Max
White Point: 0.111
Black Point. 0.111
Gamma: 1.00
Temperature: 0.0
Saturation Highlights: 0.0
Saturation Shadows: 0.0
Micro-smoothing 2.0
Highlights Smoothness: 0.0
Shadows Smoothness: 0.0
Shadows Clipping: 0.0
 
The only things I really vary here are 'Color Saturation', 'White Point', ' Black Point', and 'Micro-smoothing'.  People say to adjust the white point and black point until the the flat portion of the bell curve at both ends are gone, but I really adjust it until I think the shot looks most natural.  I also check the difference between 'High' and 'Max' smoothing and pick the better of the two.  Sometimes I use High and sometimes I use Max.  Just depends on the shot. Color Saturation does what it says and just lets you put more or less color into your final product by adjusting the slider.  The micro-smoothing does a good job of removing halos and getting rid of some noise by moving the slider to the right and increasing the value, but it will also makes the shot duller the higher the number you go.  So, I usually keep it low and let Noiseware take care of it later on. I really don't mess with the other sliders.  Save the file as an 8-bit tiff (you can use 16-bit, but I want to save space on my hard drive).

3. Mask if needed. This portion may need to be done more than once depending on how many areas need to be corrected.  I usually do the sky first, especially if it's a night shot and then I will mask in people/objects after that (I got this trick from Trey Ratcliff's HDR tutorial and modified it slightly):

     A. Open _tonemapped file in Photoshop CS4. Here is where I mask in or out anything that really didn't turn out too well during the merge process.  So, if I have people that are ghosted, a noisy sky, or anything else that doesn't look nice, I go find one of my raw brackets that looks the closest and the best for what I would like to see in my final shot.  How I do this is first, copy the background layer (CTL-J) and then layer the raw bracket in between the two layers. Open the raw bracket that I wish to mask in, hit CTL-A, CTL-C and then go back to the tonemapped shot and hit CTL-V.  That will put the bracket at the top of the layer stack.  With the cursor, left click and hold that layer and drag it in-between the background layer and the copy.  Now, create a layer mask by highlighting the top layer, then go to Layer | Layer Mask | Reveal All.  With the brush tool, use one of the round ones and set the opacity somewhere around 50%.  Make sure the white mask thumbnail is selected on the top layer and make sure the slider on the color palette is all the way to the right so that the color is back.  Now, on the picture on the top layer, I'll start painting out the areas of the tonemapped shot that I would like to see corrected.  Around the edges that I want to keep, I vary the size of my brush and the opacity of it to make sure they blend properly.  In areas that are not close to the edges, I usually go 100% opacity to make the job go a little quicker.

B.  Once I have all the areas masked in, I adjust the bracketed raw layer to make it looks more like the tonemapped layer.  To do this, I usually adjust the brightness/contrast and hue/saturation.  Click on the the middle layer and open Image | Adjustments | Hue/Saturation and adjust to get it the masked in portion to look as close to the tonemapped portion as possible.  Once complete with that, open Image | Adjustments | Brightness/Contrast and adjust until it starts to blend with the tonemapped layer until it looks like it's part of the picture.  Usually I will have to adjust these down quite a bit, but it will all come back later in the Nik Color Efex portion of the workflow.

C. Once I'm done with this, I merge all the layers by right clicking on a layer, select Merge Visible. Then I'll restart the process again if I have other areas to mask in.

4. After masking comes the part I really enjoy and that is the Nik Color Efex portion:

A. I use the Nik Color Efex (Nik CE) plugin for Photoshop (PS).  I think the stand alone works similarly to the plugin, but it may be a pain to keep going back and forth between PS and Nik CE.  The first filter I go for is the Pro Contrast filter.  Open Nik and go to Pro Contrast and click OK.  It will create a layer on top of the original background layer.  Sometimes Nik can go a little too far when adjusting the colors so I like to keep some of the original colors and tones in it.  So I usually start by adjusting the opacity to 67% generally.  This is just a good reference point that's pleasing to the eye.  Depending on the shot, I may add more or less opacity.  Again, 67% is a good general starting point.  Next, I use Tonal Contrast the same way as the Pro Contrast.  I go about 67% with that as well.  Everything begins to look a little noisy there, but that will get cleaned up later on.  The last filter I use is Bleach Bypass, and I blend that in at about 10-15%.  Again, this is generally speaking.

B. Once this is all done, I will merge the visible layers by right clicking on the layer and selecting Merge Visible.

5. Now it's time to remove noise and sharpen it up:

A. Here I use Imagenomic's Noiseware Pro.  It installs as a plugin and works really well.  So, create a layer by hitting CTL-J.  Then go to Filter | Imagenomic | Noiseware.  I mainly use either the Default or the Weaker Noise setting.  If there's not too much noise and a lot of detail, go with the Weaker Noise setting.  The thing is phenomenal.  Sometimes I will even adjust the opacity of the layer to keep a little more detail.  After this, merge visible layers again.

B. Now to sharpen it up.  Nik Sharpener looks and works similarly to Color Efex.  I just use Output Sharpener and the defaults on that portion.  It will add another layer, but it's a little funky when it's done because the sharpener layer sometimes goes away when it's done, even though the image will be sharpened.  So, I usually go to Edit | Step Backwards in PS to get that layer back and then adjust the opacity down to around 85%, generally speaking again.

Now here, sometimes I'll mask out the sky with the previous layer because I usually like the really smooth look that Noiseware gives to skies, whether they are clouds or clear.  When Nik sharpens them, it can give the sky a sort of jpg "blockiness" for lack of a better term.  So, I use the same technique discussed above for masking in people or objects, only I do it with the sky here.

Merge the visible layers once more.

6.  Lastly, I'll do something I like to call "lifting the mist":

What I do is, I adjust the exposure in Photoshop ever so slightly.  I create a layer (CTL-J) and then go to Image | Adjustments | Exposure.  The settings I generally use are as follows:

Exposure: +0.08
Offset: -0.0020
Gamma: .96

These are all a general starting point for me and sometimes I'll adjust the opacity of the layer if it's too much.  Mainly this darkens the darks a little bit and brightens the brights just a hair.  Really, it darkens more than brightens and your mileage may vary.  But, once I do this, my shots usually get some added depth to them, like you could almost jump into them.  Be advised, you may lose some detail in your darks and some may not like that, but I prefer the overall look to the shot instead of having the slight detail in the dark areas.  To me, it looks as if the fog or mist has been lifted, hence my little term "lift the mist".

Merge the layers.

7.  I then usually save the file as a tif and then save another copy as a jpg for posting.

That's it!!! Thanks for reading and let me know what you think!  If you try it all or a portion of my work-flow, please let me know how it worked for you.  I'm really curious to see if others get similar results.

17 comments:

  1. Ill be trying it shortly, but I have CS3 and Topaz. Thanks for the details!

    ReplyDelete
  2. I hope it works for you. Would you mind letting me know what you think? Thanks.

    ReplyDelete
  3. Hey Evan, I have to say Thanks so so so much! To be honest I have been struggling to find the right workflow and yours works Incredibly! Very Detailed and explained very well.

    ReplyDelete
  4. Thanks Jake! I'm glad it works for you. I was a little nervous to know whether or not it would work for anyone other than me, but now I'm beginning to feel better about it. Thanks for taking the time to comment. I appreciate it!

    ReplyDelete
  5. Thank you Evan!! I Just finished my first photo with your workflow and I am so grateful for this post, btw I merged it with CS5 and still didnt retain any Exif data, I was hoping that it would but oh well still worked great!!

    ReplyDelete
  6. Thanks for the info on CS5. That's a shame, but I was afraid that would happen. I'm really glad this worked for you and I hope you'll let me see what you've done sometime! BTW... You orb shots are really cool.

    ReplyDelete
  7. I tried merging 9 images to HDR in CS3...my computer did not like that very much, even with 4gb ram ddr3... I like the "lift the mist" but i also changed it around a little bit depending.

    As for photomatix, I created a tonemap profile, called "EvanGearing" and click it just to see what it looks like. I ran a new set of 8 images through photomatix, with your style and mine. then I blended them at about 55% mine. And i have to say, it is probably one of my top 5 favorite photos of all time. (I still have 600 photos to get through from my trip in the southwest though...) Its not really awe inspiring, but the details and colors turned out pretty good. I'll post it up on my flickr in the next week or two.

    On the negative side, it definitely (obviously) doesn't work for every shot. I have some of horseshoe bend that I tried, and they just looked horrid. I think this is mostly due to how saturated the colors there to begin with. Each set still needs its own specialized version of tonemaping... :)

    thanks for the info!

    ReplyDelete
  8. You're welcome Greg. I'm glad something here helped, if not too much. Yup, each shot is gonna be different and these are just general guidelines I use.

    ReplyDelete
  9. Hi, Evan...I'm really glad I stumbled upon your blog. I'm going with my family to Washington D.C. for the 4th of July and the following week. I've been gaining more and more interest in photography over the last year and a half or so. I purchased my first D-SLR, a D90, about a month ago. I'm REALLY hoping to get some shots that are at least half as good as yours. Thanks so much for this tutorial...absolutely fantastic work!

    ReplyDelete
  10. Thanks Ryan and good luck in DC. I imagine you will get some great shots with the fireworks there! I'll look forward to seeing them and let me know if the tutorial works out for you.

    ReplyDelete
  11. Evan this is very helpful and I have found that playing around with my own technique it is close to yours...one note / question for you...in section 3A when you do masking etc do you select all layers including the background and run an auto alignment in CS4 first...I do and it helps often with ghosting lines around objects...then I start my masking with layers like you / Trey has taught us...I do my hdr merge work in photomatix so maybe that's the difference..I have not tried the CS4 hdr merge.....Steve

    ReplyDelete
  12. Yes, I align all layers. I guess I need to add that in there. I like the CS4 merge because I think it aligns the initial hdr better than photomatix does, but I think photomatix is getting better. Try merging in CS4 and then tone-mapping in photomatix and see how that works for you. One thing you might notice is that the colors coming out of the merge in CS4 are a little less saturated sometimes, but you can always bring that back. Thanks for stopping by!

    ReplyDelete
  13. wow! thank you so much for the tutorial. the nik plug-ins don't work for me in cs5. do you have any experience with those plug-ins on cs5?

    ReplyDelete
  14. No, but I think Nik has updated them to work with CS5. I only have CS4. I think in some of my posts I may have said that I use CS5, but I think those were Freudian slips on my part. Wishful thinking I guess. Sorry.

    ReplyDelete
  15. Save as an OpenEXR - it results in an image with close to twice as much color and luminosity info than a .hdr file plus it keeps the EXIF data intact. The .hdr format is ancient.

    ReplyDelete
  16. Thanks anonymous! I'll give that a try!

    ReplyDelete
  17. Thanks a lot for sharing, very detailed, like the tutorial.

    ReplyDelete

Disqus for Evan's Expo