Commons:Photography critiques/January 2015

Photo Authoring

Hi guys,

Anybody knows who is the author of this photo?I intend to use in my website.

Thanks

File:Railay_Beach_Panorama.jpeg

@Jpdero: You'll find all the necessary information here. Additional information about using pictures from Commons/Wikipedia on your website can be found here. Please feel free to ask any further questions that may arise at the Help desk. Hope that helps, --El Grafo (talk) 20:18, 10 January 2015 (UTC)

This section was archived on a request by: El Grafo (talk) 14:10, 14 January 2015 (UTC)

Brindled sky

Hello, colleagues!

About a month ago I nominated this photo for QI. It wasn’t promoted by reason of “brindled sky” (and for some other reasons, but they are out of the question now). Really, the sky is brindled... The photo was processed (sharpness, contrast...), but the “brindles” can be seen on the initial photo, too... At that time I didn’t pay special attention on this problem, but now, when I was looking over pictures on my computer, I noticed the similar brindles on other photos. All of them were taken under similar conditions: clear sky, ISO 100...200, the exposition 1/100 or less, f=11 or less.

What is a reason for the appearance of such brindles on the sky?

Is it a phenomenon of, so to say, “the noise nature” (as I supposed first)? H’m, but it seems to me the situation isn’t especially “noisy”... What then?.. :-(

I will be very obliged to you, if you provide me with some tips. What is the way to avoid the brindles when a photo is taken? Can the brindles be removed with the processing? What is the main idea of such processing?

Thanks in advance, Dmitry Ivanov (talk) 21:45, 8 January 2015 (UTC)

Hello Dmitry, I loaded your image into gimp and played around a little bit. An indirect conclusion is that the sensor must have shown some noise, even at these relatively low ASA values of 100 or 200. But then some filtering must have happened, probably some median filtering with threshold (guessing from the result). The result is that there are these groups of pixels with same colour. Technically due to the ironing out of individual pixel noise one looses the effect of dithering. Due to the removal of high spatial frequencies one instead gets these lower-spatial-frequency effect, these brindles. It does not help that cameras record only into 8 bit per colour, but that is a contributing issue only. And it is not a JPG compression issue, as there are no artifacts visible that hint to too low a quality setting.

Of course the brindles effect can get amplified in post-processing. But then you say it is already in the original image...

Then I looked at your camera name. It says NIKON. I do own a NIKON COOLPIX S6600. With that camera I also observe issues similar to your brindles. It looks like image surfaces are painted, with loss of fine detail within patches. Even on the best quality settings the Nikon camera compares badly with the Canon bridge camera I now mostly use, and the resulting JPG file sizes of the Nikon could easily accomodate much better image quality.

To rule out sensor unevenness, one would need several photos of blue sky. If the structure were exactly the same, then the sensor response would be inhomogenous. But my bet is on sensor noise.

What can you do? You could look whether your camera can shoot in RAW (and with this I really mean unprocessed image data directly stored to card). If not, you cannot do much. You could filter further, with brindles disappearing but bands appearing. But the information you would actually need has been filtered away. You probably would need to switch to a different camera manufacturer, as Nikon has a known track record of damaging image data by design, one example being that in one study a Nikon camera was assessed unusable for astronomical images.

It may sound harsh but this is an honest assessment. Hope your next camera will be better. -- KlausFoehl (talk) 14:43, 9 January 2015 (UTC)

Just a note: The manipulations that Nikon performs that make the images unsuitable for astro imagery aren't usually relevant in normal photography, certainly not in daylight. I agree with the previous assessment that it's probably a problematic interaction of noise, noise reduction, post processing in the camera - probably with some help from compression. Shooting in RAW should fix it and helps to improve the quality anyway. If that's not an option, you might have to look deeper into the image processing option of the camera. — Julian H.^✈(talk/files) 18:00, 9 January 2015 (UTC)

According to the EXIF data a en:Nikon D3100 was used, so shooting RAW (here: NEF) should be possible. If you want to continue shooting JPG, I'd suggest to try if switching the the noise reduction setting changes something. --El Grafo (talk) 08:55, 10 January 2015 (UTC)

Thank you, dear colleagues, for your detailed explanations and opinions, they were very useful and gave me some ideas. I looked over photo forums and found opinions (not universally recognized, though) that the noise reduction on Nikon reflexes sometimes works incorrectly: some of photographers advise to deactivate the noise reduction (it is activated on my camera). As well there are an opinion about the poor work of the standard mode of the post processing in the camera and advises to use specialized modes.

So I need some experiments; unfortunately, they cannot be carried out immediately; it is winter and the clear days are rare :-(. And I will think about the work with RAW-files (you are right, El Grafo, Nikon D3100 records them).

Dmitry Ivanov (talk) 17:14, 10 January 2015 (UTC)

One more thing. You can decrease the relative noise contribution if you "expose to the right". -- KlausFoehl (talk) 14:34, 14 January 2015 (UTC)

I read about this method but never used it purposefully. About 2 weeks ago, before my first question here, I took some photos, which were intentionaly overexposed a bit, + 1EV, (sunny day, comparatively dark objects on the snow - I preferred details in objects). Really, I got rather good pictures, and snow is wholly satisfactory, but the noise on the sky can be revealed even on an unprocessed picture. Dmitry Ivanov (talk) 21:03, 18 January 2015 (UTC)

"Exposing to the right" as I understand does not mean to overexpose. It means that that brightest parts of the photo (let's assume numerical values 0..255) have values close to but still smaller than 255. If you look at the intensity histograms for red, green, and blue, you want to have a small gap, but a gap between content and right edge. -- KlausFoehl (talk) 12:32, 30 January 2015 (UTC)

Devices with LEDs

• 

This one was very difficult to me because the LEDs had completely blown the red channel, making them look almost entirely yellow. I had a hard time trying to recover them in post, and I'm still not quite content with the results. I probably should have made multiple exposures and merge them in post, but I didn't notice that at the time of shooting. Also, in hindsight, I'm wondering why I chose that ridiculously slow shutter speed of 3/10. Anyways:

• Do the LEDs look plausible to you?

Any other comments are of course welcome as well. Cheers, --El Grafo (talk) 09:51, 15 January 2015 (UTC)

El Grafo Not at all :D, what happen here? We have 3 different colours of red LED, could you share the original? Than I could try to fix, or you can take the second in the bottom (I added a note) and then clone stamp it, cause this one is good. -- RTA 10:01, 30 January 2015 (UTC)

Thanks for being honest, RTA! Yeah, looking at it after some time again makes me realize they look pretty awful. What happened there is basically that I completely overexposed the LEDs – mainly in the red channel, making them look yellow (where only the red channel was blown) to white (where all channels were blown). Here's how I tried to fix this, as far as I can remember it: I did one regular RAW development for the whole device and one for the LEDs. In that one, I pulled the exposure down as far as possible and shifted the color towards red. So everything that was white before would now be plain red. Then I stacked the two developments in GIMP with the "normal" on on top and made that one transparent at the locations of the LEDs, letting the other red layer shine through there (sparing out dots to get the highlights from the "normal" layer). That worked very well for the "11" in the display panel, which was also affected by this to some degree. However, I'm not too experienced in GIMPing around on a pixel level, so I've left some ugly artifacts in other places.

For the record: The top left LED (next to the "33 amps") actually is a different kind of LED (brighter, can have different colors) than the other ones. Also, the big one at the top right is of much lower intensity than the rest – this one didn't need any retouching at all, so I used this as a guide and tried to match the other ones to it.

I suppose someone with better retouching skills could make it look credible, but at the moment I'm really considering re-shooting the whole thing: One exposure for the whole device and a separate one for the LEDs would eliminate the need for messing around with blown channels. I'd also like to try a third exposure for the glow of that vacuum tube, which is far too dim. After all having a real tube in a digital modeler is one the major selling points of this device …

--El Grafo (talk) 11:46, 30 January 2015 (UTC)

El Grafo, if you want, send me the RAW file to rodrigo.argentongmail.com, and I could try to solve. :)

And I could fake the lights on, from here: File:VOX ToneLab ST.jpg :P

And just one note, I can see a brighter light in the left, if you gonna take another photo, have that in mind. Maybe Evan-Amos could help in this type of photo. -- RTA 11:57, 30 January 2015 (UTC)

File:Concattedrale di San Giovenale.jpg

Just to not pollute the Commons FP, I bring here some thoughts about the photo File:Concattedrale di San Giovenale.jpg also see Commons:Featured picture candidates/File:Concattedrale di San Giovenale.jpg

Livioandronico2013, I really like the texture in the floor, I like the colours, ..., this image have a better composition then this one File:Cappella della Madonna Del Rosario in St. Mary above Minerva.jpg, the celling in churches are important part of the architecture, but ok... again panorama ;). But the lack of quality remains in another points, as you did not used the sharpest f-stop of your lens, that is ~f/4.0 [1], and in this case, is quite vital cause this lens have a good area of sharpness, and a really bad result in the rest :D . Did you locked the mirror up to take the shoot? And use a remote control or a timer to take the shoot? This kind of thing avoid micro vibrations in photo after you press the shot bottom. And if the f/5.6 did not cover the DOF needed you could focus bracketing also, and tilt a little bit. Maybe Diliff could give more tips :). -- RTA 10:49, 30 January 2015 (UTC)

It is a pity that the composition cuts the central arch. -- KlausFoehl (talk) 11:56, 4 February 2015 (UTC)