[Blurrr]

Hi, please let me explain – and I apologise if this has already been discussed (I did look). I know that a digital photo is made up of pixels that are arranged in a regular – regimented pattern. It is this regimented pattern that – for me – gives digital photos that digital look. I can’t really describe it but there is something very distinctive about them. My question is – is there a way to ‘jumble up’ the pixels to make the photo look more film like. Or perhaps there is a filter in PhotoShop that does the same thing?

[shuttertalk]

Hm... I know what you mean, but not sure what I can recommend. Maybe try one of the blur algorithms, or perhaps adding some noise?

[guerito]

There are Photoshop Actions that claim to make your picture look as if it were taken on film, even resembling the characteristics of a film brand (i.e. Velvia), see here.

I don't know if this answers your question. Regarding the pixels, I always thought if you have enough of them you can't tell the difference.

[loopy]

Mmmmmm.... Velvia...

Velvia is the only thing I miss about Digital

[slejhamer]

Blurrr, can you post an example of what you are seeing that you don't like? And some info about your camera type, the file size you are using (e.g., "JPEG Small", or "RAW", etc.), and how you are processing them? I think we can be more helpful if we have more info.

[Kombisaurus]

Hi Blurr,

Some of the issues I can think of that affect digital images to make them look different to film include:

* Jpeg Artifacts - poor or agressive file compression (especially in web graphics) tend to do nasty things with nice gradations and can put halos around high-contrast edges. This is usually a problem associated with post-processing where the user cranks up the compression when saving an image for the internet, but can also be noticable with some cameras when they are operating in their highest compression/lowest quality settings. It can be addressed by using less compression (ie better quality settings that mean larger files). Better still, use RAW images if you camera supports them.

* Sharpening Artifacts - A lot of digital cameras (and software used in post-processing) use functions to try to sharpen an image after it is taken. While this can be useful, if done too aggressively it can leave halos around high-contrast edges and give sharp edges as nasty jagged appearance. Usually there are settings on the camera (and in software) to set the level of this sharpening, or turn it off completely.

* Noise Reduction Artifacts - This is similar to the Sharpening Artifacts issue, except these artifacts are the unwanted side-effects of aggressive noise-reduction. Many cameras and post-production software offer functions to cut down the amount of noise in the images they produce, but when this is done too aggressively it can soften an image and reduce detail as well, giving things a slightly out of focus feel or sometimes make it look a little as though you are looking at the image through a translucent curtain. Extreme examples can blur the image and make it look out of focus. Usually this can be "fixed" by reducing or switching off the noise reduction, but doing this often results in higher noise levels as well.

* Image Noise - This is the digital equivalent of film grain, and in most cases can be thought of in the same way. If you take photos with a low ISO of say 50 (which is the equivalent of using a film speed of 50) then you will get very little noise (or grain), but if you take photos with an ISO of 400, 800, or 1600 then you will get a lot more noise (just as you would get more grain in film). The main differences are that digital noise looks slightly different to film grain, and that the amount of noise at 100 ISO on one camera can be quite different to 100 ISO on another camera. The ISO value relates to how much light is required to expose the photo (and this is constant from camera to camera) - and while increasing the value will almost always increase noise, it is not a direct measurement of noise.

* Dynamic Range - although this doesn't relate directly to your mention of the "regimented pattern of pixels", it most certainly contributes to the "digital look" of photos and can often be the biggest giveaway of a digital photo (compared to a scanned film photo) in my opinion.
Most digital cameras still do not have the same dynamic range (ability to record detail in extreme highlights and extreme shadows) as good film. More importantly I think is the fact that film's responsiveness to light has a "knee" and "shoulder" which means that while it gives good contrast and a fairly linear response to midtones and mild highlights and shadows, it's response tapers off in extreme shadows and highlights - in other words it manages to keep some detail in these areas, but sacrifices some contrast to do so (when viewed on a graph, these bends in the line resemble a knee and shoulder, hence the terms). Digital sensors on the other hand have a completely linear response to light, which means they record things with constant contrast right up to the point where they run out of "headroom", at which point the signal is clipped and those areas of the photo become completely "blown out" and no additional detail can be recorded (ie those areas appear as pure white or black). This is most notable with bright skies, where digital photos are notorious for resulting in white skies when the image is exposed for a darker foreground.
This difference between digital and film can be addressed mainly by being aware of it and exposing more for the sky (or other potential problem area). Also working with RAW images allows you to "save" areas that might be blown out or severly underexposed, as the raw image contains more dynamic range than the finished jpg image and allows you to non-destructively alter a number of the shooting parameters after the shot. Another technique I use when I'm using a tripod and dealing with a scene that has extreme contrast (ie dark foreground and bright sky) is to take two identical photos, one exposed for the light areas and one exposed for the dark, and then merge them on the PC later to get the good bits from both images. This technique can mimic the way film responds, as you are sacrificing contrast for detail.

* Channel Clipping - This is an extension of the Dynamic Range issue, but this applies usually when only one "channel" (the red, green, and blue component of each pixel) runs out of dynamic range and is clipped, but other channels are able to record detail. The end result is a distortion of the hue of the colour. Sunset photos are notorious for this, and it will appear as a slightly different coloured red/orange ring around the sun. This usually goes by without being noticed, but is often a givaway of a digital image. If you brighten really dark shadows you often see the result of this as well, as the brightened areas may have exaggerated and unrealistic hues (although they will usually look fine when left at their original brightness).

* Bayer Pattern Distortion - This should only be an issue with fairly poor quality cameras or low-resolution cameras, but although virtually invisible, is still present in almost all digital cameras (except for a few that use tri-layer sensors). Basically each pixel on the camera's sensor is only sensitive to either red, green, or blue light. The image that the camera first gets is therefore a monochrome image, but each pixel is responding to a different colour. The camera can then interpolate the colours based on the brightness and colours of the surrounding pixels. (Do a google search for "bayer pattern" to read more about this). If the software on the camera isn't that good, or if the optical quality or resolution aren't that good, then this can also affect areas containing fine details, especially if they contain different colours as well as tones.
There is very little that can be done to prevent this effect. Newer, better cameras with higher resolutions and better inbuilt software do a much better job of resolving the image, but the only real solution is one that is only available in a select few cameras. These have a special three-layer sensor that has seperate sensors for red, green, and blue that are stacked on top of each other. The downside of this is that a 10 megapixel camera effectively only produces a 3.2 megapixel image, as each pixel requires three sensor pixels to read it. I can't comment on how well these work, I just know they exist.

I hope that helps. I'm sure there are other differences (ie purple fringing, etc) but these are the main ones I could think of.

Cheers
Adrian

[shuttertalk]

.... yeah, what he said.

Thanks for the awesome essay, Kombi!

[Blurrr]

Thanks very much for the replies - esp Kombisaurus who has really nailed things. I suspect it is a combination of these things that I am seeing. Knowing this and the limitations of the CCD will help me greatly. I am only using a Canon A75 - it doesn't have a live histogram but if I view the histogram of the photo taken then I will be able to adjust accordingly to get as much of the dynamic range of the CCD.

Something Kombisaurus mentioned intrigued me:

"Basically each pixel on the camera's sensor is only sensitive to either red, green, or blue light. The image that the camera first gets is therefore a monochrome image, but each pixel is responding to a different colour. The camera can then interpolate the colours based on the brightness and colours of the surrounding pixels. (Do a google search for "bayer pattern" to read more about this)."

Something I have been interested in recently is the talk of a dedicated monochrome CCD and what benefits it would have over greyscale conversion in PhotoShop. If the image the camera gets is first a monochrome image, is this the image we get when we set a camera to B&W setting. Is this a better/more accurate/higher quality image than the one that is created by the colour interpolation? I am trying to get my head around this but I guess it takes three pixels (RGB) in combination to create a colour. So if the pixels were not displaying colour you would get an increase in definition? Completely wrong I'm sure - but I'm off to read about bayer patterns now.

[Blurrr]

Sorry I have gone off topic but I thought I would finish my B&W thought. Found a good review of the discontinued Kodak DCS-760m:

http://www.luminous-landscape.com/review...760m.shtml

Where the reviewer states:

"Without an anti aliasing filter and no Bayer color matrix, the resolution of a 6 mega pixel monochrome camera is astonishing. In monochrome, 6 mega pixels effectively does what it takes 12-24 mega pixels with a color matrix."

[Kombisaurus]

You hit the nail on the head Blurr,

(interesting article at luminous landscape too)

The tri-layer sensor cameras are a bit interesting (I think it might have been Kodak that developed the sensor? and perhaps Fuji or Pentax that use them in a camera? hmm.. not sure).
I'm not sure how sharp their colour images would be, as I'd imagine it would be quite difficult to have the three sensor layers lined up exactly and so on... but I'd imagine you could get fantastic monochrome images from them by just using the red, green, or blue channel as the monochrome image, if you don't mind throwing 2/3 of your pixels away

Hopefully this technology has a real future, and we'll be seeing 3x16mp cameras and so on down the track.

Cheers
Adrian