Bringing light to the dark
The digital dark room is not that different from an old style film dark room. Many of the tools used are the same as the days of film, in the digital dark room we can straighten an image, crop it, apply dodge and burn, adjust the exposure and change colours.
Some people get very uppity about post complaining that it is not real photography if you shoot a picture and then edit it, others say that post is vital (go and speak to a fashion photographer or their model!). The important thing to remember is that you can never fix a bad photo in post and that professional photographers have been using post processing techniques since the dawn of photography to actually make their images. The renowned photographer Ansel Adams was well known to make extensive use of post processing techniques to make his final images and always urged photographers to 'Expose for the shadows, develop for the highlights' showing that he clearly made use of post to great effect.
A raw nerve
Before I start talking about post I need to explain a little about image formats. Most of you are hopefully aware that most digital cameras save their images as JPEG files (short for Joint Photographic Expert Group) and shown by a .jpg or .jpeg on the end of the file name. JPEG images have a nasty little secret, in order to record what the camera sees and keep the image file to a small size a JPEG does not keep all the information captured by the camera's sensor, it discards some of it. For this reason JPEG is called a lossy compression system, that is to say it compresses the information captured by your camera but in doing so it loses some of that information.
Now-a-days most SLR cameras will also shoot in a mode called raw, which has various file prefixes, for example my 7D adds .cr2 to the end of the file name. The difference in file formats is that a raw image contains all the information your camera sensor gathers, it is a lossless format.
But why am I telling you all this I hear you cry? The answer is quite simple, when you change an image in post certain actions will loose data, if you've already lost data when your camera has compressed the image to a JPEG format then the output will get worse as you make big changes to the image or if you make several small changes.
To illustrate this imagine that this sentence is my picture:
I've compressed it by removing the spaces and punctuation. You can still read it perfectly well but it's not as easy to read.
Hello world isn't it a lovely day!
As a raw image the sentence is the same, nothing changes but as a JPEG it becomes:
Helloworldisntitalovelyday
Let's say I now do some post on it, for example adjusting the exposure (more on this later) I'll represent this by removing the o's, in raw we get:
We can still kind of understand it, but what about the JPEG:
You can see how this can rapidly lead to a JPEG becoming such a poor quality image after manipulation (note this is a very extreme example it's not quite this bad when talking pictures because we have much much more data).
The trade off here is that to make a JPEG your camera will make certain modifications to make the image look better and so raws will often need some sort of post doing on them and raws contain a lot more information and so are much much bigger than JPEGs. The advantage of raws is that you can do a lot more to them including things like resetting white balance values (see later tutorials) and make more and bigger changes in post before the image becomes unusable.
In this blog I will not go into detail about what causes this as it is way over the top. The important take home is, if you want to edit then shoot in a raw format or, if you don't want to have massive image files or your camera doesn't shoot in raw (perhaps because you're using a phone or compact) then before you edit the image convert it to a .tiff (Tagged Image File Format) which is another lossless compression method, edit this file and then save back to JPEG.
Hell wrld isn't it a lvely day!
We can still kind of understand it, but what about the JPEG:
Hellwrldisntitalvelyday
You can see how this can rapidly lead to a JPEG becoming such a poor quality image after manipulation (note this is a very extreme example it's not quite this bad when talking pictures because we have much much more data).
The trade off here is that to make a JPEG your camera will make certain modifications to make the image look better and so raws will often need some sort of post doing on them and raws contain a lot more information and so are much much bigger than JPEGs. The advantage of raws is that you can do a lot more to them including things like resetting white balance values (see later tutorials) and make more and bigger changes in post before the image becomes unusable.
In this blog I will not go into detail about what causes this as it is way over the top. The important take home is, if you want to edit then shoot in a raw format or, if you don't want to have massive image files or your camera doesn't shoot in raw (perhaps because you're using a phone or compact) then before you edit the image convert it to a .tiff (Tagged Image File Format) which is another lossless compression method, edit this file and then save back to JPEG.
Danger!
Now before I get into the rest of the blog I have a few of important points to make:
- No matter what you do post cannot make a bad image good!
- Post can make a good image bad if you don't take care!
- Post can be computationally expensive so unless you have a speedy computer don't try to watch a DVD or do other computationally challenging tasks at the same time as trying to run some post!
Grand tour
First off is to open the GIMP and get the photo into the editor. Once you have opened the GIMP you can get a photo into it in two ways; you can do this either by finding it in your 'My Computer' ('Finder' for Mac users) and drag and drop it into the main screen or you can go to 'File > Open' under the menus as for any normal programme.
So lets run through what you may want to do to a picture. My example that I will be using is the one below. Now I'm not so happy with the horizon, its not level, or the exposure, its under exposed, or how flat the grass seems, its not very bright and interesting, so I will be showing you how to straighten and crop an image, alter the exposure and alter the brightness of a specific colour in this tutorial.
Now if you have a look at the above picture there are a few things to show you. The main window in the middle is where the picture that we will be working on will be displayed. Either side of this are two toolboxes. The one on the left has the main editing tools in it, things like selection tools, text, clone/stamp, straighten etc. The one on the right has the brush selector at the bottom and above that a space for viewing layers, channels, paths and the undo history. Today we will only be considering layers (which looks like three grey squares in a stack) so make sure that this is selected. The undo tab in this toolbox is useful as it allows you to select something you have done and undo do it or redo it.
Sometimes using a tool out of the editing tool box will pop up a window, as will selecting some tools from the tool bar. These windows can then be moved around to let you see what you are actually doing.
Stack 'em high
The first thing that I do when bringing an image into the Gimp or Photoshop is to create a new layer, now for the modifications that I am going to show you today this is not that important, but for some other modifications (like getting rid of spots) it is very important. If you don't want to get confused with the layers you can miss this step out for now. One thing it is useful for however is to flick back and forth between the layers to see what your original image looks like and what the modified version looks like.
To create a new layer for working in right click on the main background layer which is seen on the right hand side of the screen in the layers toolbox. Then select 'Duplicate Layer':
This creates a second layer in which we will do all our work. Make sure that it is the one that is highlighted and that it has an eye next to it so you can see it. The GIMP can, sometimes confusingly, do things to a layer that is not visible; this is because it decides which layer to do things to not by which is visible but by which is selected, just because there is an eye next to a layer showing it is visible does not mean it is the one that will be worked on, make sure to always have the layer you want to work on selected i.e. it is highlighted (blue on my screen)!
Twist and shout
Now the first task I did was to straighten the image. To do this choose the rotate tool from the editing toolbox (shown in a slightly darker grey):
This opens a pop up window like the one shown below and overlays a grid over the image (ostensibly the grid is meant to help but I don't think it does). In this window you can change the angle by typing in the 'Angle' box, or, using the up and down arrows or using the slider bar. You can also rotate by clicking in the image (somewhere towards one edge or other) and dragging your mouse up or down. You can also, if desired, change the position around which you rotate the image.
So I played around with the angle for a few minutes until I was fairly content that the horizon was level. Note in this screen shot you can see the background layer has not been rotated highlighting the point I made about making sure you have the working layer selected!
Let there be light
Now I have a straight horizon I want to brighten the image a bit and tone down some of the really bright bits. To do this I have elected to use the curves tool which is found in the 'Colors' menu (note the American spelling, grrr).
This opens another pop-up window that looks something like this (note the shadow grey background curve will look different for your pictures).
Now what does this window tell me? Well starting at the top it tells me I'm being naughty and ignoring my own advice and working in JPEG (but will skip over that)! Below that I have the option to select from a preset curve setting that I have saved before (beyond this tutorial). Below that is the channel selector. This has 5 settings: Value, Red, Green, Blue and Alpha. We will only look at Value and Red, Green and Blue today. You can also choose here to look at the graph on a linear or log scale. Then onto the meat of this window, the graph. The grey curve in the background shows the distribution of light and dark in the image from ALL colours. If the image was all black then there would be a single line at the left hand side of the graph going from the bottom of the graph to the top of the graph, if the image was all white then there would be a single line at the right hand edge of the graph going from the bottom to the top. There isn't either of these lines, which is great, this means my photo is not completely over exposed nor completely under exposed. What the graph does show however is that I do possibly have a bit of under exposure. A quick check of the image will confirm that, it looks very dark!
Now if you look at the graph you can see that there is a grey line running from the bottom left corner to the top right corner. This is the normal curve, what I want to do is tweak this curve to make the shadows brighter and the bright bits darker but maintaining white as white. But what does all that mean!
The first thing to understand is that I can move the line, selecting curve type 'Smooth' will help me here as it tries to keep things simple, you can select 'Freehand' and go way out with lots of changes but let's not do that. So what does changing the curve do?
The y-axis of the graph shows a graduated scale from black at the bottom to white at the top. If you look at the point at which the diagonal line gets to the y-axis in the bottom corner you will see that it is at the blackest of this graduated line. In other words black is black! At the top right you can see the opposite the normal line for white is at the white end of the graduated scale, white is white! Now, if I was to drag the normal end point at the bottom left hand corner up the scale black won't be black anymore, it will start to get whiter, i.e. it becomes grey. Conversely if I drag the white end down it starts to get blacker. See where I'm going?
What this means is we can adjust the exposure of an image and simultaneously make areas that are too dark brighter and areas that are to bright darker! Much better than in the camera where we can only make everything brighter or darker!
Got that? Great if not try it for yourself and then come back and join me to discuss the changes that I have made.
So, I have an underexposed image, but it also has some over exposed bits. Now, I want to keep white as white so I won't touch the very top right hand corner but what I will do is pull the bright areas back a bit. To do this requires a bit of guess work but pick a bit on the normal curve where it is bright, left click on the curve and drag it downwards, looking at your picture at the same time (make sure you have preview selected and be patient, even with a speedy computer it may take a while to re-draw the image). You'll see a dip appear in the normal curve, the black line will separate from the grey straight line like mine has and the bright bits of the image get darker. Halfway there now on this! Now for the dark bits; again guess where you want to make the change at the left hand side of the curve and drag it upwards, looking at your picture. You'll see the dark bits get brighter! If you make a mistake don't worry just hit the 'Reset' button and try again. Once you're happy hit the 'OK' button and admire your handiwork, you have now successfully rescued a slightly dodgy exposure!
Green green grass
I like the exposure now but the grass seems a bit dull. There's two ways I could change this, one is by using the Hue/Saturation method the other is by using curves again. Hue/Saturation allows you to change the colour of a colour (i.e. make a green more bluey or more red etc.) and change how vivid the colour is (more saturation make the colour more vibrant, less saturation makes the colour more grey). I'm going to leave Hue/Saturation for another day and play with the curves again. I opened a new curves window from 'Color > Curves' again and this time selected 'Green' as the channel. If you were doing something where you wanted a blue or red to be brighter you'd select their channels.
This gives you something looking like this:
As an aside you can see why it is a bad idea to do post on a JPEG. Look at the big bit of the background curve at the left hand edge. See the big white lines? Now compare that to the curve that I used when adjusting the 'Values' channel. Those white lines show where I have lost information. If I do too strong a change or too many changes to a JPEG the histogram (the technical name for the grey background curve) will get more and more white lines in and the image quality will get worse (see why I ignored my advice ;)).
This window works in a similar fashion as the 'Values' channel except instead of making colours brighter dragging the curve upwards makes it more green and downwards more black. I guessed what level I wanted to be more green and made my adjustment, but oh no look what happened!
Everything became more green! That's not what I wanted to happen. Never fear though, hit cancel or reset to get rid of the curves window and go back to normal! to make this change only affect the grass I need to make a selection to tell the GIMP what to make this change to.
To do this I'm going to use the 'Select by colour' tool (top right in the editing toolbox which looks like a hand pointing at a stack of coloured squares). This selects things based on them being a similar colour to what I point at. I clicked on this selection tool and then clicked on the grass in the image, you can hold down shift to make multiple selections and add them together. Once I was finished I got this:
This looks hideous and slows the computer down no end (see why you can't watch a DVD or similar at the same time as editing your photos!) but shows you what the computer has selected for you. I then went back to the 'Curves' editor via 'Color > Curves', selected the green channel and made my adjustment. Now, instead of changing the whole picture it only changed the grass to make it more green, giving me this:
Now my grass definitely looks better and its time for the final finishing touches!
Cut down to size
I can't leave the picture like this, it's no good. The problem is the rotation creates blank areas which the computer cannot fill in. The grey chequerboard you can see in the corners shows the areas where it is missing bits (I hid the background layer by clicking on the eye next to it). It's also not a great composition as there are a couple of bits of grass sticking up on the left. To fix this I need to crop the image (note some editing suites may have a 'Rotate and Crop' tool which will do enough of a crop that you have a rectangular image of the same proportion as your original, but if you want to cut anything else out then you will still need to manually crop).
The crop tool looks like a scalpel (3rd row, 4th icon). To use it select it and then left click and hold and drag across your image to draw a rectangle. In the options at the bottom of the edit toolbox you can select specific aspect ratios, turn on guides and specify positions and sizes for the crop by using the text boxes. For this edit I just free handed it, judging my start and stop points to roughly give me 1/3rd sky above the boat and to take out the annoying grass on the right hand edge. Giving me a final image of this (note I have added a watermark by following the instruction in the watermarking tutorial):
Become something
All this is great I now have an image I'm happy with but I still can't use it yet. The problem is that the GIMP, like Photoshop, creates its own bespoke image files made up of all its layers. To do anything useful with the image I first have to make it have only one layer and I also have to export it as a usable image file format.
First off is the flattening. I have three layers in the image, the original image, the layer I have been working in and the text layer where I added my watermark. To flatten it go to 'Image > Flatten Image' and this will squash all the layers down on top of each other to give me one picture. Now for advanced users you can do funky things with layers when you flatten but that is for a different time for now GIMP's default setting (at the top of the layers toolbox) of 'Normal' should work fine.
If you intend to put the image up onto the web to speed up transfers and reduce the amount of space your image uses it's important to reduce the image to a more manageable size. If you're not intending on putting it online you can ignore adding the water mark and this resizing step. To resize the image go to 'Image > Scale Image':
This will bring up a pop up window which looks like this:
For now leave the quality and resolution settings alone and concentrate on the Image Size 'Width' and 'Height' settings. Make sure the drop down box has 'px' in it for pixels and that (unless you really want to change the aspect ratio) then chain is not broken and enter a size value for the longest edge. I often use a value of 1024 pixels for the longest edge. The computer will now work out the value it needs for the shortest edge to maintain the current aspect ratio. Hit scale and all of a sudden your image will become much smaller:
So now I have an image that is almost useful but it's still floating around in the GIMP's bespoke file format so I need to export it to be able to use it as an image. To do this go to 'File > Export As' which will give you this screen:
We need to pick our export file type so in the drop down box where it says 'All export images' click on the drop down and select 'JPEG image', choose where you want the file to go to then press export.
Once you've pressed export you'll get a pop-up looking something like this:
Choose your image 'Quality' (note the higher the quality the bigger the file) and hopefully the GIMP will tell you an estimate of the file size (for some reason it didn't for me this time!), hit export and you will get a new JPEG file in the location you wanted it all edited and ready for sharing on the web (or if you missed out the scaling and watermarking phase) for printing or displaying on a computer etc.
Post post
So what I have shown you is the basic steps of adjusting a poorly exposed image and lifting a colour to make it more vibrant and interesting. It may seem a bit daunting when you first step into the digital dark room, I'm an engineer and use computers all day everyday but even I sometimes struggle to make things look right or remember how to do things (especially in the GIMP now I've moved to Aperture!) but stick with it and practice and you'll get there in the end!
I intend to do more tutorials on specific aspects of post, and if demand is there do some face to face stuff using the projector in the community centre. If this is something you would like to see then make your voice heard on our Facebook group or at the next meeting which is scheduled for 9th September 2014 at 1930. If you have any questions about what I have discussed in this tutorial give me a shout on Facebook or at that meeting!
The biggest thing to remember with post is that it cannot make a bad image good, no matter how hard you try, concentrate on shooting good images and then just tweak them in post and you'll be the best you can be! Also remember that post is as much about personal opinions as the rest of photography, there are several ways to achieve the same objective and none of them are necessarily right.
The final closing word I would have on the subject of post is don't get dispirited when you have a bad day in the dark room. Keep practising and see what you can get it to do, don't over do it either and think you have to do post on every image, concentrate on fixing ones you really like and want to keep. Keep an eye out for other tutorials on it on here and ask fellow group members for help!
Over all have fun, remember it's a hobby and happy snapping!






















