Episode 062: Noise in the Dark!

Download the Video!
Download the companion file! (…if there is one…. ;-) )

I am through a bad cold and nearly through an iris inflammation and so I loved to talk – you get a double dose of MTG tonight. Philippe helped me out last week – and he will be a regular contributor after that success.

After I solve the mystery behind the menue entry “Edit/Fade” I tackle an old image. Taken 4 years ago with my old camera (Nikon D70) at ISO 1600 and f=1/1.8. It’s noisy.

I use the noise reduction from UFRaw and combine two results of the RAW conversion – one with and one without noise – to one. As with the selective sharpening I left the edges of structures noisy and kept the noise out of the structureless areas. Of course this is done with layers and masks. with this technique you keep the structures crisp and the noise out.

After finishing this video, I found out that UFRaw has a well hidden option for removal of chroma noise. More next week.

This image still needs some work on it – I’ll show that in the next episode.

Joel has a lot to say about noise and I’ll go into some other noise reduction techniques in the future.

The TOC:

03:15 Edit/Fade
09:55 The image in UFRaw
14:45 Noise
16:40 Noise reduction in UFRaw
19:10 Comparing the images
22:50 The Plan
23:35 Making a layer mask with “Edge Detect”
29:40 Recap
30:40 Fine tuning with another layer
38:50 Rotating
41:50 Cropping


Creative Commons License

This work is licensed under a
Creative Commons Attribution-Share Alike 2.0 Germany License.

14 thoughts on “Episode 062: Noise in the Dark!

  1. Hi,
    because of the small size of the video and the high compression I saw not to much of that noise, but the technique is clear and really helpfull. Very nice picture by the way, I like it very much.
    The difference between the two versions of the picture is strange. Maybe there is an explanation for it. What do you think about setting the blendmode to Difference instead of reducing the opacity of the upper layer for fixing this?
    I like such videos, which are not showing a single function of Gimp but the hole process of the editing.

  2. I’ll put some snippets in 100% into the posting tomorrow.

    I always forget about the difference mode…. should watch some instructional videos. ;-)

    The image is one of my favorites – perhaps I can make a “final version”this time.

    I think I’ll do a complete image walkthrough more often in the future. I like them too – but the image has to have a special challenge.

  3. Rolf, your remark about a “final” version reminds me of a (yet) another request to submit for some future show. That is, how does a serious photographer avoid spending inordinate amounts of time tweaking things — or at least, how does s/he restrict such an investiment to the pictures that deserve special attention.

    In one video I especially enjoyed, you showed spending 5 minutes each on “snapshot” processing .. I think what I envision is showing an example (for “serious” or perhaps work-related photos) of workflow and a mental-xray of the process of sorting promising images, identifying required adjustments, and then finding the right amount of “playing” while applying the personalized final touches.

    I suppose this question invites a range of answers from people with different work styles and creative/analytical coordinates, eh? But, still, it might be nice to hear things like: ‘in this photo, I perceived something “special” about the composition (or lighting or colors or textures or ..)’, or ‘I thiought this might be good for illustrating such-and-such a technique because …’

    Hope this makes /some/ sense. :-)
    ..jim

  4. I simply have to wait with answering this question until I have made the step from “Amateur” to “Photographer”. But this podcasting leaves me no time to make progress. Wait – I have not made that much progress in the decades before that…. ;-)

    I don’t know an answer for that. Some images just jump on me – most when I take them. I then already feel that I have a “keeper”. Rarely I find something while looking through the images of a shoot that says something to me.

    Sometimes these keepers loose their appeal during post processing, sometimes they get shot down by the critique of others. And sometimes I know that I want a different result than the one I have achived.

    Perhaps I’ll throw this question to the Focus Ring guys in one of the next talks. We’ll record one on October 2nd, but I already have a topic for that. And a bit more thinking time…. ;-)

  5. @jim
    Maybe the videos called “Light Diary” from Craig Tanner on RadiantVista are the stuff you search, if you want the hole process from the beginning “why did I make that picture” to the end.

  6. @Fomit

    Hey, thanks for the link. That is pretty interesting stuff. I watched one episode (#8) and even learned some self-help psychology tips!

    I’ll have to try some more of them.

  7. Hi Rolf,

    first of all, thanks a lot for your interesting videos.

    I downloaded the mtg62.zip file, and opened with gimp the xcf file.. well, I am not sure if it is due to different monitor calibration.. but the image is almost full of strong red noise (expecially in the foliage). Enabling or disabling the “edges” layer, doesn’t change things so much.

    However, I tried to open the .nef file with ufraw, and it is much more noisless if compared with the image open in gimp. I don’t know where the difference comes from (either the 8-bit conversion, or the camera color profile set inside ufraw). Any suggestion?

    Regards

    Marco

  8. Well, nice show again! But I am absolutely not happy with the denoising of UFRaw. As you pointed out, the image becomes washed out, almost as watercolours. A lot of work is needed to correct this (as you showed) and still you can see the denoising. I have way better experience with GREYstoration, which has an edge preservation algorithm.
    As a second trick I often denoise only one channel. For instance as you showed the red channel has the most (almost the only) noise. So why not decomposing the image, denoising the red channel and composing it again??!! This works fine for me in many pics and is much less work than the edge detection stuff you did. Maybe you can drop a hint in the next show…

  9. If you use a windows-system, you could export the image without denoising and use Noiseware. If you are not satisfied with the result, you could do the same as Rolf did with one layer with noise and the Noiseware-image.

  10. It may be a bit late to post to this thread but someone may stumble up on it and find this idea interesting, there is a way of using 2 or more shots called image averaging to reduce noise in images. It involves changing the opacity of the layer with each layer being a copy of the image so they let so much of the layer below through until you reconstruct the image. So for example the percent of the original layer being 100% layer 2: 50% layer 3: 33% layer 4: 25% and so on.

    The main use may be when shooting lon exposures your camera may have a lot of noise so you could shoot at shorter shutter speeds with a higher ISO and use image averaging to reduce colour or sensor noise.

  11. Pingback: MTG: Обзоры выпусков за весь 2008 год! « Блог фотолюбителя :)

Anything to add from your side of the computer?