Confusing colours

[Nicoleica]

Advertisement (gone after registration)

I wonder if anyone can help explain this to me? Whilst testing out my new Elmar 24/3.8 today, I noticed strange colours on white/bright areas of the DNG. I'm quite happy to accept that these areas are over-exposed, but what has confused me is that they are most noticeable in looking at the DNG with Aperture. Looking at the same DNG with Capture one shows a vastly reduced incidence, and looking at the JPG-Fine from the camera (Using Preview.) hardly shows any colour oddity.

 

I've attached 100% crops from the views from each of these 3 programs. (1st Aperture (DNG), 2nd Capture One (DNG) , 3rd Preview (JPG))

 

Any help would be greatly appreciated.

 

Thanks,

 

Nicole.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[Nicoleica]

Just to make things even more confuzzled, I've just looked at the same DNG file with Photoshop Elements, and it hardly shows any hint of odd colours.

 

All of these examples have been without any alteration. It seems strange that 3 different programs should show the same image so very differently. Or maybe it's just me?

 

Attached screen shot from Photoshop Elements.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[farnz]

Nicole,

 

I assume that you're referring to the front edge of the guttering and the windows sills? If so, then it looks to me like a form of 'blooming', which can occur where highlights are clipped. I commonly get it as a red halo round burned out streetlights during night shooting and is associated with over-saturation of the pixels at that point.

 

Presumably different raw developing software uses different algorithms, which might account for why some show it more than others.

 

Sometimes blooming is unavoidable, e.g. the streetlight scenario, but often it can be tamed through looking after the highlights by dialling in -1/3 or -2/3 EV. (Apologies if this sounds like teaching you to suck eggs or if I'm way off the mark.)

 

Pete.

[Nicoleica]

Thanks Pete,

 

I hadn't ever noticed it before on my photographs, and was at first alarmed (In case my Imp had picked up a new bad habit.), and then confused as 3 different programs showed very different results, and the JPG showed a better result than the DNG.

 

Being ancient, I rarely do any work to my photographs on the computer, apart from cropping. I still try do everything in the camera, before I release the shutter. I've got a lot to learn about digital manipulations, so all advice is more than welcome.

[jaapv]

Nicole, it just means that C4 handles the files better than Aperture, which is not surprising. If you use C4 you can do quite a few things, in the tab "Q" you can adjust exposure, colour temperature ( click on a spot that should be neutral using the eyedropper tool) you can adjust contrast, saturation etc and you can "tone down" the highlights and shadows.

There is a sharpening, cropping and straightening tool two tabs to the right and the whole program is rather intuitive. Converting it into Photoshop Elements is sufficient for resizing and preparing for print or web. I would advise you to go that way for best results. Look on it as your darkroom. As a metaphor, Jepgs are prints you get from a 24 hour lab, DNG converted ones are the ones you create yourself.

[sandymc]

Usually when you see something like that in a blown highlight, it's usually a side effect of the conversion between different color spaces.

 

To get a bit more technical, in camera, a blow highlight is 255,255,255. However, in any non-camera color space (e.g., sRGB, Adobe RGB), that translates into something else, usually with one color channel blown, say 240,230,255. So you can get color offsets that can be quite difficult to deal with. Using C1 or Aperture, you're first translating into an internal space (e.g., for Aperture its Wide Gamut), adjusting exposure in that space, etc, etc, then converting to another color space for display.

 

Ironically, this is usually less of a problem if you are converting into relatively narrow color spaces, as shown in this case by the in-camera JPEG and Elements, neither of which would use the very wide color spaces that Aperture or C1 are using internally. But there will also be differences just because of the different color rendering of C1 vs Aperture. And different demosaicing algorithms will also interpolate around a blown highlight in different ways.

 

 

Sandy

[Nicoleica]

Advertisement (gone after registration)

Thanks Jaap & Sandy,

 

I'm glad to hear that it's not a new fault with my camera. (Phew!) I can see that I'll have to remember the limitations of 8bit colour spaces, and learn how to get the best from them. Up until now I've mainly used Aperture for organising and printing my photographs. Now I think I shall have to learn the difficult bits too.

[jaapv]

It is better, Nicole, to stay with 16-bit colour spaces for a large part of the post-processing process. Although Sandy is right (he always is ) on the artefact reduction by converting to a small colur space, actually doing contrast corrections like curves and levels can lead to severe loss of tonal values and posterizing in an 8-bits environment. (before any myths kick in, this is something completely different from the bit-depth the camera produces!) So if you have Elements 6, do set your DNG conversion to output a 16-bits file (the blue link in the middle under the preview. It should be set to " aRGB 16-bits ") The same with C4, export an aRGB 16-bits file to PSE6.

[Nicoleica]

Thanks again Jaap. I only have Elements 4 at present, and it doesn't seem to want to upgrade past that. (Elements came free with my Digilux 1 a long long time ago, and until yesterday, I hadn't used it in years.) I'll probably try and learn what I can do through Aperture first, as that enables me to print directly, which Capture One doesn't.

 

Thanks again to everyone for your help and advice.

[Nicoleica]

Jaap. I've had a play today in Aperture using the the 'dodge' tool, and it helped a great deal. Thanks again for the advice.

 

I did notice one strange thing though. I noticed from the Exix data, that the DNGs were using Adobe RGB, even though the colour space set on the camera is sRGB. JPG files show sRGB, but DNGs show as Adobe RGB no matter what the camera is set to. Is this normal, have I gone bonkers, or is there something silly that I'm doing wrong? (Probably all three!)

 

Thanks,

[jaapv]

No when you are shooting RAW, the camera setting is irrelevant. Your DNG converter is setting your colour space to aRGB, which is the better colourspace for editing. You should do your Aperture work in 16-bits aRGB, and only convert down to 8-bit sRGB for web publishing (or some simple printers). Your best output for printing is usually aRGB 8-bits.Jpeg, or for high-level printing 8-bit TIFF aRGB. This presumes that you have a decent monitor that is calibrated and can render aRGB.

[Nicoleica]

Thanks Jaap. I've learned a lot in the past 36 hours.

[Nicoleica]

For anyone who may be interested, or has had this same problem. I have discovered today that it seems to have been compounded by the default 'RAW fine tuning' in Aperture.

 

The two screen shots below illustrate this.

 

The first picture is 'as is' after importing into Aperture. The 'RAW fine tuning' setting defaults to '2.0' and the unexpected colours are very apparent.

 

The second picture shows the same area after setting the 'RAW fine tuning' to '1.0'. The unexpected colours are vastly improved.

 

Thanks again to everyone for their help and advice on this matter.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Aldophoto]

Very interesting! I've noticed that also the histogram changed a bit.

Besides, have you compared this last result with C1?

As an Aperture user, I'm still wondering like Hamlet: C1 or not C1?

 

Aldo

[Nicoleica]

Hi Aldo,

 

I haven't compared too many photographs yet, but from those that I have, I would say that Aperture gives as good a result as C1 using the 1.0 raw fine tuning setting. I prefer to use Aperture if I can as I've been using it for several years now, and have photographs from several cameras in my library. Also, I can print directly from Aperture, which I cannot do from C1.

[adan]

As to the original problem - it is color moiré or aliasing.

 

The thin white edges of the gutters/windows are slightly sloped relative to the "grid" or "checkerboard" of pixels on the sensor, and thus are alternately cutting across red/green and blue/green pixel pairs, which creates the orange and cyan barberpole stripes.

 

Perhaps more accurately, the thin white line alternately is NOT captured by blue or red pixels, persuading the camera that it is a candy-colored line, rather than white.

 

Evidence of the missing anti-alias filter in the M8.

 

M8 jpegs apply some electronic fuzzing or antialiasing, which explains why the effect is less noticeable in the jpeg.

 

RAW developers also vary in their native ability to correct for moire/aliasing, which is why C1 seems to do better. There may be a moire/control slider available in Aperture (I wouldn't know - I just just Adobe Camera Raw).

 

If there is no specific tool for correcting this, often it can be reduced by playing with the Color Noise control, which basically blurs the color information without blurring the luminance (B&W detail) image.

[Nicoleica]

Thanks Andy. Now I understand why this mainly seems to affect thin straight lines. This also explains why I hadn't noticed it before with other cameras. Perhaps this an argument for 'honeycomb' pattern sensor grids? (Either that or a need for fatter photons!) And thanks to everyone's help, I've been able to find a useable solution to it too.