Wikipedia:Featured picture candidates/VADM John Bulkeley 1988.jpg
- Reason
- I think this is a good portrait of a United States Navy Admiral and Medal of Honor recipient
- Articles this image appears in
- John D. Bulkeley, List of United States Naval Academy alumni (Medal of Honor), Board of Inspection and Survey
- Creator
- Robert Lucier. Not a WP User
- Support as nominator --Kumioko (talk) 00:23, 18 July 2009 (UTC)
- Oppose unfortunately, it's full of artifacts, probably caused by oversharpening.
- It's not full of artifacts, it's a (very nicely) scanned film-based shot showing a bit of film grain. Noise reduction, IMO, would ruin it. If it is to be judged fairly, it shouldn't be examined the way we do digital shots. --mikaultalk 07:33, 18 July 2009 (UTC)
- It does have jpeg artifacts, but whoever it was that opposed was probably referring to the noise, I concede. I don't think you can categorically say that noise reduction would ruin it though. It would certainly ruin detail if strong NR was applied across the entire image. Or more to the point, in this image it would accentuate the jpeg artifacts which are buried in the noise. Using a bit of intelligence and only performing noise reduction on areas without texture minimises the detrimental effect of NR as per Alt 1. Diliff | (Talk) (Contribs) 12:04, 18 July 2009 (UTC)
- Hey, whatever turns you on ;-) you've just demonstrated exactly what I was on about though, for me it's now a manipulated derivative of its former self and just looks weird, all that gritty detail giving way to selective smooth bits. The background looks good but maybe algorithm-based NR would look a little less "tweaked". Perversely enough the alt's margins and transitional areas now look slightly artifacted to my eyes. Ho hum. --mikaultalk 02:29, 19 July 2009 (UTC)
- You're right that there is some artifacting around the 'transitional' areas as I had feathered the selection before running NR - perhaps I feathered it a little too much, but a sharp transition would potentially look worse. The thing is, like I said, all the artifacts were there to begin with - you just chose to ignore them because it was old-school film and consistently artifacted across the whole frame. ;-) When you say algorithm-based NR, what exactly do you mean? The noise reduction software analyses/profiles the image before performing NR on it, so in that sense it uses an algorithm. Diliff | (Talk) (Contribs) 07:58, 19 July 2009 (UTC)
- It's the masking I guess. I dare say the software you used can be tweaked to identify and ignore fine detail and it's that I was referring to. IMO you're generally ok with chromatic stuff and (digital) luminance noise but it's nowhere near so good on uniform film grain where the even best-tweaked algorithm is going to have to be so aggressive that detail loss is unavoidable. In any event I prefer a global correction to avoid these odd-looking (IMO) textural variations across the image. Masking digital shots is always a good idea, but with film-based stuff you should do all or nothing; if grain is bad enough to really need NR it's probably beyond redemption anyway. OTOH if there's good detail that would be lost even with extremely skilful global NR, it's probably best left as-is. --mikaultalk 12:29, 19 July 2009 (UTC)
- You're right that there is some artifacting around the 'transitional' areas as I had feathered the selection before running NR - perhaps I feathered it a little too much, but a sharp transition would potentially look worse. The thing is, like I said, all the artifacts were there to begin with - you just chose to ignore them because it was old-school film and consistently artifacted across the whole frame. ;-) When you say algorithm-based NR, what exactly do you mean? The noise reduction software analyses/profiles the image before performing NR on it, so in that sense it uses an algorithm. Diliff | (Talk) (Contribs) 07:58, 19 July 2009 (UTC)
- Hey, whatever turns you on ;-) you've just demonstrated exactly what I was on about though, for me it's now a manipulated derivative of its former self and just looks weird, all that gritty detail giving way to selective smooth bits. The background looks good but maybe algorithm-based NR would look a little less "tweaked". Perversely enough the alt's margins and transitional areas now look slightly artifacted to my eyes. Ho hum. --mikaultalk 02:29, 19 July 2009 (UTC)
- Thanks for the touch up. --Kumioko (talk) 22:09, 18 July 2009 (UTC)
- It does have jpeg artifacts, but whoever it was that opposed was probably referring to the noise, I concede. I don't think you can categorically say that noise reduction would ruin it though. It would certainly ruin detail if strong NR was applied across the entire image. Or more to the point, in this image it would accentuate the jpeg artifacts which are buried in the noise. Using a bit of intelligence and only performing noise reduction on areas without texture minimises the detrimental effect of NR as per Alt 1. Diliff | (Talk) (Contribs) 12:04, 18 July 2009 (UTC)
- It's not full of artifacts, it's a (very nicely) scanned film-based shot showing a bit of film grain. Noise reduction, IMO, would ruin it. If it is to be judged fairly, it shouldn't be examined the way we do digital shots. --mikaultalk 07:33, 18 July 2009 (UTC)
- Support (alt-1 only) - Great noise reduction by Diliff. - ☩Damërung ☩. -- 16:58, 18 July 2009 (UTC)
- Support (alt) Noodle snacks (talk) 09:11, 21 July 2009 (UTC)
- Support Alt 1. I don't have a problem with the selective NR and I think it allows you to focus on the subject instead of the noise. Diliff | (Talk) (Contribs) 10:01, 21 July 2009 (UTC)
- Support Alt 1 - Nice and crisp photo, now. Staxringold talkcontribs 13:06, 21 July 2009 (UTC)
- Oppose Whilst I have to commend the noise reduction of the edit, the face still has a little too much noise for my tastes. 03:25, 25 July 2009 (UTC) —Preceding unsigned comment added by Labrat256 (talk • contribs)
Promoted File:VADM John Bulkeley 1988 NR edit.jpg --wadester16 05:09, 25 July 2009 (UTC)
