Make photos look … retro? Digital lomography? [tech rant]

Janet and John, discussing Lomography:

Now, these are troubled times when it comes to photography (what camera to use?) – and, as a matter of fact, audio (digital or analogue?), and movies (who needs high quality when we are in love with our TV screen?).

What makes Lomo pictures what they are?

You did not come here to read my analysis report – you came here to download a Photoshop Lomo Filter. If it only were as simple as that. There is a whole range of predefined Lomo filters available for The Gimp or for Photoshop.

As you will see below – if you care to read through this at all -, these automatic scripts do not provide a full digital reproduction of Lomo image properties.

They only allow you to apply simple approximations.

If you bear with me, I will detail my own analysis of Lomo images.

You can download my own “Lomoizing” Photoshop filter and action set that shows how more precisely operating, semi-automated Lomo Photoshop filters could look like. And while you can download and install this right away, it may be helpful for you to read the instructions as well.

 

Being a cautious digital photo editor, you would not just apply any automatic script filter – you would apply filters manually and judge from the image itself what filters would be appropriate. “Judge yourself” is my colloquial way of wording it – really, I mean “look at the image, contrast features, histograms, and analyze what you see”.

Depending on your specific photo, peripheral blurring, peripheral contrast change, color shifts, application of noise and peripheral level range adjustment may be more or less appropriate. You may also care for your saturation features and for your curves manually and individually. You could select some color ranges, feather the selection and desaturate or color-shift them, and enhance saturation for some other feathered color range selections. If you allow me, I’ll try to explain the differences to digital photography that I found.

While there seem to be no established hard rules, it pays to carefully implement Lomo picture features on digital images rather than running simple scripts that have the potential to turn your image into something cheap looking.

Now, I am not just such a fan of the “Lomo camera” per se – I just like some aspects of these pictures, particularly the warm colors. I believe that we have ‘gone digital’ for a reason: to increase our degrees of freedom. Digital cameras do not only ‘still rule’ ‘despite the Lomo’ – they rule even more: With a perfect digital picture, you can do a lot more than just Lomo’ize it. On the other hand, the Lomo has some coolness – but not for all the money in the world, because you are stuck in one way of photography – and it is in no way the only ‘cool’ camera around. We shall see about that in a minute.

The reverse is really a problem. It can be really cumbersome to ‘de-Lomo’ize’ an image; you would have to start with white balance and then worry about the colors next. If you want crisp fast analogue photography and look for better alternatives to a real Lomo, any Olympus “mju” or Yashica pocket camera will likely give you a far more robust experience at a far lower price than Lomo cameras; also, second hand mirror reflex cameras can be purchased at fair prices. It is not hard to find a camera that is cheaper than the Lomo, easier to carry, and that shoots sharper pictures (as the Lomo optics seems to blur shots very easily). Many point and shoot digital cameras will include color shifts, noise and vignetting – so all you have to do with such images is shift colors. Anyway, detailed results of my camera evaluation are here.

Nevertheless, I find that the Lomo take on ‘design groove’ is very cool, and I like some of the overdrive of the intensity of these pictures for many types of photography – except scientific documentation, or other image recording that requires high quality.

1. Vignetting and exposure

As you can easily see, the following two pictures (first: regular version of digital camera picture; second: Lomo’ized version) somewhat resemble the real Lomographics – pictures taken with real Lomo cameras: contrast and saturation are increased, and the margins and the center are off in terms of lighting.

The Lomo Kompakt Automat allows for rather long exposure times (unlike most other compact cameras) so you can get some blurry glary shots. Of course, if all you do is take blurry and glary shots, there is a risk that this gets boring real fast. The lens of a Lomo causes vignetting (light fall off in the corners.. darkening), which is an interesting feature to study and implement digitally, as contrast, color saturation and focus vary compared to the center of the images. Digitally, vignetting and bright blurs are rather simple to implement by overlaying your picture with gradients or changes to the peripheral margin of the image.

 

 

2. Noise

I analysed a bunch of real Lomo photos and digital photos with regard to the noise. The following two examples show an original Lomo image (a), a blurred version of the same image (b), and the difference (c). What you see is uniform noise distribution which is, in fact, a typical Lomo Kompakt Automat feature, and partly, a feature of point and shoot digital cameras; I did not find nearly as much noise in the digital mirror reflex camera (Canon EOS 300D).

The following two examples show an original Lomo image (a), a blurred version of the same image (b), and the difference (c). What you see is uniform noise distribution which is, in fact, a typical Lomo Kompakt Automat feature, and partly, a feature of point and shoot digital cameras; I did not find nearly as much noise in the digital mirror reflex camera (Canon EOS 300D).

 

 

You can easily see that there is a rather uniform distribution of RGB-noise predominantly present in dark and midtone areas of the image, but not so much in very bright areas of the picture.

This ‘feature’ easily illlustrates that Lomo cameras do, in fact, produce technically “inferior” images.

At any rate, such noise can easily be applied to digital pictures, by employing a feathered selection to dark areas of an image, and by adding noise specifically to these dark areas in a way that they fade out into the bright areas.

A good way to try that out would be by playing with this idea in your favorite image software – manually.

Furthermore, most Lomo pictures are far less focussed than one would hope. If you really need to, you can blur your digital images, but I will definitely not implement blurring as a ‘key concept’ in Lomo’izing photography. There are enough blurry pictures on this planet, enough said.

3. Color shift

The multicoating on the Lomo Kompakt Automat lens apparently allows for some vivid saturation of some colors, and for some color shift of others.

So if we want to achieve Lomo like images in digital photography, we need to understand these differences in detail.

Looking at specifically taken authentic Lomo pictures, a first analysis will reveal that the typical distribution of RGB-information is shifted compared to digital images. Keep in mind to always pay attention to the weight: is the image warm, or cold? Why is that? Is the white really white, and the gray really gray?

If you are after fast solutions, you have to understand that in many Lomo images, Red image information is a nuance brighter, followed by Green image information, and there is still brighter Green information than Blue information – compared to how most digital cameras store it. This is only an overall image assessment, and not true for each single color.

Comparing the Lomo histograms against some typical digital photography histograms (in my example: Canon Ixus v2), you will easily discern a global red shift. Lomo image brightness has a discrete shift to the red – which may explain their warm (rather than red) appearance.

 

 

The following pictures are typical examples of Lomo pictures. Despite the weather being bad, despite the sky being gray, and despite the color not appearing ‘reddish’, the shift of the curves present in authentic Lomo images is rather obvious once one directs attention to it.

 

 

 

 

(The last image shows histograms only pertaining to the rectangular area selected; this is to show that a red shift is present throughout the ‘non-red’ areas of the image. As this pictures contained some redlights, I decided to avoid analyzing these, as they hardly provide insights into any underlying red-shift of the Lomo technique).

Now to the details. Comparing overall histograms is only a very rough approach to color shifts caused by lens multicoating.

I took a number of images side by side, with a Lomo Compact Automat and a Digital Ixus camera. When I measured RGB-space differences of discrete color ranges on benchmark object photographs, I found that Lomo images feature specific color shifts as detailed in the following diagram.

The diagram lists different colors in digital images and the changes one needs to apply to them in order to get the correlated color in the Lomo picture. The colors have been grouped according to the change vector in RGB-space. As you can easily see, color changes are not uniform! So, for bright gray colors in your initial digital image, you would shift the initial RGB values (200,200,196) by (-6, -3, -10) in order to get the adjusted Lomo values of (194, 197, 186). For greenish yellow (listed as ‘yellow 3’), you would shift the initial color RGB values of your digital image from (138, 167, 45) by a displacement of (10, 26, 28) to obtain the resulting Lomo value of (148, 193, 73). And instead of just transposing one particular, precise color (say, bright gray [200,200,196]), a whole color range of related colors could be transposed by a given color shift (say, bright gray to

And instead of just transposing one particular, precise color (say, bright gray [200,200,196]), a whole color range of related colors could be transposed by a given color shift (say, bright gray to white range [190-250,190-250,190-250]). Such can be achieved by increasing the “fuzziness” on a selection of a color range.

The colors have been grouped according to the change vector in RGB-space. As you can easily see, color changes are not uniform! So, for bright gray colors in your initial digital image, you would shift the initial RGB values (200,200,196) by (-6, -3, -10) in order to get the adjusted Lomo values of (194, 197, 186). For greenish yellow (listed as ‘yellow 3’), you would shift the initial color RGB values of your digital image from (138, 167, 45) by a displacement of (10, 26, 28) to obtain the resulting Lomo value of (148, 193, 73). And instead of just transposing one particular, precise color (say, bright gray [200,200,196]), a whole color range of related colors could be transposed by a given color shift (say, bright gray to

And instead of just transposing one particular, precise color (say, bright gray [200,200,196]), a whole color range of related colors could be transposed by a given color shift (say, bright gray to white range [190-250,190-250,190-250]). Such can be achieved by increasing the “fuzziness” on a selection of a color range.

The diagram data shows grouped color vectors (ward’s algorithm, not standardized, multivariate grouping by displacement RGB vector).

 

If you care to implement something like this yourself, feel free to use the listed values from this diagram. There are two ways to implement this in Adobe Photoshop CS: one way is, to use the ‘Image > Adjustments > Match Color’ command, the other way is to apply color shifts directly. Both are now available as presets in my Lomo-filter-set (download my Lomo package for Photoshop ).

In order to use the ‘Match Color’ command, you will need color statistics files (.sta – files). I have collected all of my real Lomo Kompakt Automat images, and I have collected a bunch of Lomowall-Images from the Internet, and I produced color statistics files from both. They are part of the download package.

When applying such color statistics using the ‘Match Color’ command, you will obviously have to load a statistics file first. You then should exert extreme caution to make sure that you use the slider controls ‘Luminance’, ‘Color Intensity’ and ‘Fade’ to your advantage – and not just hit the ‘ok’ button.

I recommend a ‘Fade’ setting between 30 and 70 which ends up blending the predefined color palette of real Lomo pictures with the color palette of your digital picture. I also recommend to make sure your colors are not oversaturated. In fact, many real Lomo pictures are not oversaturated at all.

If you apply my ‘Shift Color Vectors’ filter, you may get different results compared to matching color statistics, so you may try both and see which one you like better.

Whatever you do, keep in mind that no Lomo picture has histogram pixel buildup in extreme histogram areas – i.e., there should not be large amounts of pixels distributed in the very dark (color picker values around 0), or bright (color picker values around 255) histogram regions.

Instead of stark black or white images, the typical Lomo white is typically a greenish grayish bright color, and Lomo black is a noisy dark grey. So depending on what your initial digital image shows – and digital images can indeed have lots of white or black -, you may have to use the curve tool in order to move all image signal to the middle range of, say, 30-200 (rather than 0-255).

4. Geometry

Without going into further detail at this writing, Lomo pictures do distort geometry compared to regular digital cameras due to their wide angle objective.

This can be accounted for – to some degree, at least – using the ‘spherize’ command in Photoshop. I have attempted to code that into my downloadable photoshop action set.

5. Applying digital Lomo-izing tricks

The following images show examples for applying different concepts of “Lomo properties” as observed in Lomo pictures, to a digital image.

All filters were applied ‘manually’ – no ‘script’ was defined, as the steps are easily reproduced with a little practice. For the following example, some black vignetting was applied, the color tones were shifted to a red predominance as compared to the digital ‘original’ that featured bluish highlights.

Last but not least, darker areas were selected with a feathering border, and a small amount of Gaussian noise was added. A full appreciation requires the full resolution images; you should be able to obtain these by clicking on the pictures below (but be patient, that may take a while). Please realize that the examples here are somewhat exaggerated. A more subtle and careful application of the observed differences would not provide such a stark contrast to the original digital image.

So if you think that’s cool …. do you need to buy a Lomo right away? The answer is hinging on your view on labels, on your view on what a camera should represent, and on your view on other options.

We are living in a world that is stuck on ‘labels’. If I tell people that I applied Lomo filters to digital images, and if these filters are not carefully tweaked, people tend to not like the results. They will look for “mistakes” or “inaccuracies”. I can tell you right away that this is the most sophisticated digital Lomo filter currently available for free download, and talk about image features in all detail. But if I tell people nothing about any filtering, and if the filters are applied carefully, manually, and with diligence, people typically love the results – and no explanations are required.

There is a weird notion of ‘authenticity’ around these days that entirely ignores the fact that Lomo cameras are simple, cheap, physical constructions that will tweak the light the same way every time a picture is taken – which is exactly what good digital filtering does. Also, the fact is ignored that digital cameras often represent cheap constructions that will repeatedly cause images to feature distorted colors – such as strong blue or green overtones – to a point where correcting these images is mandatory.

But such is the reality: one can not offer results per se – results must be labelled. If you believe in ‘true’ labels, you must have that Lomo. If you can afford to invest research, a certain amount of mouse clicks involved in image tweaking, and the time necessary to come up with decent digital enhancement, you can easily get by using a digital camera.

Unless you are required to provide clean scientific photo documentation (upon which you should probably think about getting a digital mirror reflex camera), you’re fine filtering your photographic material anyway, no matter how. And while we are at it, I may add that my favorite little camera (Canon Digital Ixus) already adds a tad bit of vignetting and noise all by itself.

The most difficult part for you to answer is the question, what a camera would represent to you. There is no straight answer.

The Lomo standard issue camera is a cute little thing, that makes people smile when they see it. It has these little red lamps that indicate that the button is about to be pushed. It is really hard not to laugh when the Lomo is aimed at you. It feels nice, it looks nice, and there is some funny class to it. Group photography and candid photography require a small camera, that is cute and not intimidating. So a Lomo may just do the trick after all.

If you photograph in social settings, you will get some frightened looks on people’s faces by waving a big mirror reflex camera with a big objective on it. You may get some nice staged photography – but surprise shots depend on some socially acceptable form factor. So you want to seriously consider something cute and small.

Of course, Lomo is not the only small camera, and nice group shots have been done with all kinds of other cameras. If you try to find out, what a camera means to you, and what camera makes the world smile at you, you probably will follow the right path.

The following example shows an original digital image (top image), a full application of Lomo-type alterations using various digital filters as described above (middle image), and a very carefully chosen set of changes that were applied to match this particular picture (bottom image).

If you just press ‘fast forward’ on simple Lomo filter options, the result tends to always look like a “digitally edited” picture and not “realistic”. If you take time and use gentle amounts of filtering (rather than full percentages), you may end up with a much more pleasing result.

 

 

 

 

---

6. More sophisticated Lomo filtering

The following pictures were taken with an affordable point and shoot digital camera (Canon Digital Ixus v2).

Then, filtering techniques such as the ones described on this page were applied manually. These images provide examples for the results of subtle digital “Lomo filtering”:

 

7. Preprogrammed Photoshop actions to approach true Lomo feel addition to digital images

After having stated the restrictions that apply to any automated filter that runs on your images without being able to base the filtering on a numerical analysis of the picture, it is still possible to just program some filtering that will slightly shift an image into the “right” direction.

Here, you will find a zipped archive containing a photoshop action file that contains a number of actions detailed as follows, and three image statistics files for your application to your own digital images:

1. Shift Color Vectors

This filter moves some of the color ranges to a Lomo-range. All of the change vectors displayed in the above diagram are implemented.

2. Apply Image Statistics

For this, you will need to use the enclosed Lomo Image Statistics files. You can manually run these by selecting ‘Image > Adjustments > Match Colors’ in Photoshop CS or see what happens if you hit this button. I have included three statistics files: one with a limited choice of colors based on my own Lomo images (this is very useful, as it only slightly alters color tables in a very good direction), a second one that I based on all of my current Lomo pictures taken (a very full range of color adaptation is possible with this file, but I find it better to ‘Fade’ this color table so it does not get applied full force. The third statistics file is based on a range of around 70 pictures of different Lomo Walls taken off the internet; by and large, they have a similar effect as the second file.

3. Spherize

Since Lomos produce wide angle distortions, some spherization seems appropriate to some degree. Play with the spherize command and see if you can live with the result.

4. Shift colors to the red side (red button): This action will shift the color histogram to provide a mild increase on the red shift.

5. Shift colors to the blue side (blue button): This action will shift colors to the blue side (blue button): Through a mild increase in blue shift, the histogram will provide a somewhat reversed effect from the one described above. This is probably only useful if your digital image already has a very strong red-green-shift that for some reason you with to reduce.

6. Lomoyze (orange button): You can run this filter more than once if you feel that more emphasis is required. Running it once, however, does not cause your image to be outrageously distorted. The filter will cause an increase of sharpness restricted to the luminance channel of the image, it will introduce a slight central white-yellow blur, peripheral shift in contrast and lighting, and selective color saturation changes as well as an almost imperceptible amount of noise added to dark areas. This filter attempts to process digital images in a way that matches my actual observations of real differences between Lomo and digital pictures.

8. Digital filtering generally to convey an analog feel

There is an analogy of Lomo-esque pictures being just the result of clever digital filtering, in sound: if you shove your audio output through a stiff compressor (analogy to increase in saturation and contrast in digital imaging), and through a two band parametric equalizer (analogy to peripheral cut-off and central boost, which is what Lomo filters really do), bingo: Lomo-esque sound filtering.

Funny enough, people never seem to complain about smooth application of carefully chosen sound filters to digitally encoded and stored music. Whether you do it on your amplifier, whether you run OSS3D on your iTunes mp3-player: Lomo-esque sound filtering will yield a boosted, relevant, crisp, real and authentic feel. Wear a pair of V-Mode headphones, a pair of Stax headphones, and you may agree.

9. Digital cameras, and analog alternatives, to use as cameras for Lomo-rule photography

Lomo cameras are regarded as ‘cool’ because they serve the purpose of ‘Lomo rule photography’.

I conducted a test, over a couple of weeks. With that test, I evaluated the following cameras and analysed their potential for Lomo-rule photography:

• Lomo Kompakt Automat (35mm)
• Horizon 202 Panoramic camera (35mm)
• Olympus [mju]II (35mm)
• Canon Digital Ixus v2 (digital)
• Canon EOS 300D mirror reflex (35mm)
• Nokia 3660 mobile phone (digital, built-in camera)
• EXA II (35mm)
• Russian FED 5B (35mm)
• Smena Symbol (35mm)

The detailed results are here.

In a nutshell, you may want to buy an Olympus [mju]II camera, if you want small, affordable and fast film action; you should buy a Russian Fed 5B, if you want a really cool looking camera with superb optics for little money (try ebBay); you should try a small Canon Ixus camera if you want a good quality and small digital pocket camera; but you are not limited in your choice of camera at all. Get a digital camera, get 39 or 42 mm lens adapters, your choice of Russian or other 39 or 42 mm connector lens and see where that takes you.

Any camera can be used for Lomo rules photography.

Lomo Kompakt Automat are somewhat cutsy, rather heavy, very expensive, and their optical lens is really not that good enough to warrant the hype. Other cameras result in similarly – if not better – color-shifted images at less price and with better image quality.