Contributed by Dylan (dtoh)
I hope this thread belongs in the right place because I don’t think it’s a tutorial about any one particular principle, just an explanation of how the fridge man came into being.
If you ever go to Kathmandu, staying at Thamel , the main tourist district has great advantages in terms of convenience, location and perceived safety. The good thing is though, it isn’t that big so you can wander out quite easily where I found photographing much more in keeping with normal local behaviour. I basically sat myself down with some locals on a storefront (who really didn’t show me any interest) and snapped away at anything that looked interesting. The main focus was on trying to get some images showing aspects of Kathmandu living I guess.
About the shot itself :
The following image shows the progression of the original RAW to the final result I posted in the ‘street’ section.
The first thing I do after a day’s shooting is catalogue the images in Lightroom. I tag them all with my name and a location (to make it easy to separate my images from Marianne’s). In Lightroom, the major corrections I find easier to handle here (over CS5) are:
- White balance correction (not needed for this one)
- Fill light and contrast
- Graduated filters (not needed for this one)
- Recovery of highlights
- “clarity” boost (which I didn’t use for this image)
- Vibrance (often no changes made because I find colour work better in CS5)
The image itself is a bit underexposed in the foreground and the afternoon sky was a haze and not recoverable for blown highlights so I adjusted sliders ignoring the unrecoverable stuff in mind.
I then export the file as a 16 bit TIFF file with no sharpening at 300dpi. Why 16bit instead of 8? No very good reason – andecdotally, I have found that the banding issues are less prominent when working in 16 bit then converting at the end to 8bit rather than working all in 8 bit. (I stress – this is personal anecdote and based on no ‘fact’)
The first thing in photoshop I do is duplicate the background layer in case I stuff something up royally. On this duplicate layer, I usually bring up the shadow highlight box and make corrections based on how much the image needs it (often I make no corrections). I don’t tend to adjust the sliders any more than 10% - especially for highlights in plain skies since posterising tends to occur. I have a plugin by Imagenomic called Noiseware pro which I am a big fan of. I usually start off using a preset then make some adjustments based on how much noise I want in the image.
(noiseware pro interface box – at 100%, details can just become smudges)
Next up, a levels layer (usually to boost contrast a little). Notice the horrible histogram with blown pixels on the right. I adjusted levels taking that out of consideration since there was little I could do anyway.
Next , a generic minor vibrance and saturation boost.
Next, I create a new layer in overlay mode with 50% neutral gray filling the image (these options can be achieved by alt-clicking the new layer icon at the bottom of the layers panel). Using a soft brush (shortcut B) with say 10% fill and 10% opacity, this functions as my dodge and burn layer. In this image, the main subject was too dark for my liking and some of the features in the stalls were also a little dark . So I set the brush to white at the above settings and slowly played around. Clicking the eye icon next to the image on and off is a good way to see just how much change you may not notice what you’ve done from the gradual changes at such a soft brush setting.
Next up – a whole bunch of selective colour work. In this image – I wanted the man to stand out not only with the depth of field but from the surrounds as well. The ground tones were a bit bright and dominated the scene a bit. Using the drop down menu for select colour (see image) , I selected the ground colour and adjusted the sliders so that as much ground was covered by the selection as possible but little else. Once the selection is made, marching ants appear on the screen and any new adjustment layer created will only affect this layer. The good thing about this selection technique is that CS5 does the feathering for you. From this selection, I created a new levels layer and darkened the ground. I could have done this in the dodge and burn layer but I find this method more precise.
Next up, the sky was a slight golden colour at the time but in this image, just hopelessly overexposed. I wanted to see if I could bring back any of the original colour realising that there wasn’t recoverable detail. One way to do this is to use the same selection tool and click on the sky. I then created hue/saturation layer with that selection and clicked on the ‘colourise’ button. What this will do is ‘paint’ the selection with the colour it detects. Sure you can paint it with any colour but you can run into great difficulty doing this. For instance, CS5 picked a colour in the yellow spectrum and I adjusted the lightness and saturation to give a slight golden hue – if I had changed the colour to blue and tried this, I would have ended up with bizarre looking halos on the buildings where the transition of the selection ceased. It’s a great tool but the temptation to use it unsubtly should be resisted.
At this point in time, other than sharpening, I’ll often stop the post processing with a final image here. The result from here is the 2nd image in the original panel of three.
The following steps all take advantage of different blend modes. So that all of the layers are visible in the thumbs , I flattened the image prior to proceeding but you can keep all of your layers in a super big file if you like – just merge all of the visible layers and work from that merged layer.
The next screenshot shows you the 5 duplicated layers that I create. I often don’t use all 5 but they are all there for me in case I need to. It is important with this layer stack to work from the top down.
The order of layers and roughly what I use them for (from top to bottom) :
- Luminosity &Gaussian blur
- Overlay or vivid light for a high pass sharpen layer
- Soft light (for contrast)
- Multiply (for darkening)
- Screen (for lightening)
Luminosity + Blur
The function of this layer is to create a glow effect. The Gaussian blur adjustment is to a huge amount that really makes the image look a blob. Changing the blend mode to luminosity makes it see-through to a degree and blurs the edges of the colour so the end result is varying degrees of softening and glow. If you do not want areas of the image to be affected, simply use a layer mask and brush them out with the ‘black’ brush. (eg. Portraits of great detail in old skin etc).
High pass sharpening:
This alternative method of sharpening is one I do like because you can change the blend mode of the layer for different degrees of sharpening. The radius is similar to what you would pick for unsharp masking. Using soft light /overlay /vivid light then provides increasing intensity of the effect . The unwanted result of this though, is that there might be some noise introduced to smooth areas. I usually layer mask out skies and smooth waters in this layer for landscapes. I prefer this over unsharp mask because you can see what you are sharpening before you click OK (see image below)
Soft light [contrast]
The next layers blend mode is changed to soft light. Leaving it on default settings I find results in too much contrast and oversaturation of the image. On this layer, I typically lazily address this by pressing Ctrl-U to bring up the hue/saturation dialogue box and reduce the saturation to a desired colour and increase lightness until there aren’t any crazily black or dark areas.
One good way to keep an eye on what you’ve done (other than clicking the eye button on and off) is to click the “\” key. This should bring up in red the areas of the image that have been masked out. From the image below, you can see that I didn’t want the man darkened and I didn’t want the already bright sky to be brightened more by the contrast effect.
I have come to prefer using this blend mode of darkening of dodging and burning because it is harder to ‘overdo’ the effect. The layers name comes from the effect you would get if you were to stack two slides over each other and look through it. With each slide, you’d be blocking out more and more light so the overall result is darkening and some contrast boosting too. I don’t use this layer extensively so when I create a layer mask (which on my PC is set to ‘white’ or see through) , I press Ctrl-I to invert the mask so that it is all opaque. I then use the brush tool on ‘white’ to brush in the effect I want. The pic below shows that most of the image is red when pressing “\” meaning that I’ve only used the multiply effect in small parts of the image (sky and buildings and slightly on the ground)
Think of screen as the opposite of multiply. I wanted the fridge man a little brighter and some bright bits of clothing and sign on the side so I underwent the same process as described above for the multiply layer.
Finally – once I’m happy with it all, I’m relieved to press that flatten button and change it back to 8bit and watch the file size drop from 1gig to 50mb or so !
For presenting on the web, it’s a few touches of border, watermark, repeat slight unsharp masking and brightness /contrast adjustment because I know my screen ends up producing stuff that looks darker on the web than it would at home.
I hope that was helpful and it probably went through a lot of the same processes I’ve described in the other tutorials. Please remember , this post wasn’t about telling you what to do or even what is the ‘best’ way to do it. It is simply a workflow out of many different ones, much of which I developed out of trial and error really.
Thanks for reading if you made it through!! (dtoh)