Wikipedia:Featured picture candidates/Bump map vs isosurface3.png

Limitations in bump mapping

Voting period is over. Please don't add any new votes. Voting period ends on 13 Jan 2011 at 20:57:26 (UTC)

Reason: This is a simple raytraced image illustrating a key (and therefore encyclopaedically valuable) limitation in the 3D-model texturing technique known as bump mapping. See the caption and the very detailed description on the image page itself for more, or have a look at the bump mapping article itself.; One guarded positive comment to an early version was made at Wikipedia:Picture_peer_review/Archives/Oct-Dec_2010#Bump_map_vs_isosurface which I should respond to: there are certainly better examples of raytracing available on Wikipedia (some of them are featured already), but that is not what this image is representing - it is representing a specific technique in raytracing using an arguably compelling "spot the difference" approach. There is nothing else which adequately illustrates this limitation so it is wikipedia's finest from that respect.
Articles in which this image appears: Bump mapping (an older version of the image)
FP category for this image: Wikipedia:Featured pictures/Sciences/Mathematics (since it graphically represents a mathematical modelling technique)
Creator: GDallimore (Talk)

Support as nominator --GDallimore (Talk) 20:57, 4 January 2011 (UTC)[reply]
Support Version 2. Very well done! Indubitably strong EV about isosurfaces vs bump maps. However, maybe it would be better to use a longer focal length to mitigate perspective distortion of the spheres. Purpy Pupple (talk) 21:12, 4 January 2011 (UTC)[reply]
- Oh, arse. Well spotted. Yes, I did have the camera at a pretty wide angle. Given that it's a 24-ish hour render, I won't produce a new version just yet, but will see if there are any other suggestions for improvements. GDallimore (Talk) 23:10, 4 January 2011 (UTC)[reply]
  - Actually, I can do a low quality render for comments. Will upload soon. GDallimore (Talk) 23:21, 4 January 2011 (UTC)[reply]
    - What processor are you using? Maybe there would be other Wikipedia members with vastly superior hardware who can use your POV-ray code to render better versions, since the software is free. Purpy Pupple (talk) 23:47, 4 January 2011 (UTC)[reply]
      - If anyone wants to volunteer, the code is all on the image page. Let's see what other comments there are first. GDallimore (Talk) 01:18, 5 January 2011 (UTC)[reply]
    - Also, the test render is pretty good. It seems that turning off global illumination (which seems like what you did) actually improved the image because in the original picture, the low number of GI samples caused the floor to appear cloudy. The lack of anti-aliasing and the presence of JPG artifacts is not good though. Purpy Pupple (talk) 17:23, 5 January 2011 (UTC)[reply]
      - The cloudy floor is totally intentional to give a sense of depth and a bit more interest to the image overall. It is achieved by a turbulent, light-absorbing fog, and is not a result of a radiosity/global illumination artifact - radiosity is only used minimally to lose some harsh shadows around the bottom of each sphere. If you think I should leave the fog off in a final render, let me know. Same for the depth-of-field blur: better with or without? GDallimore (Talk) 17:35, 5 January 2011 (UTC)[reply]
        Removing the fog a focal blur (which was on a really high setting) cut the render time down to 40 minutes! I think I prefer it without the fog, although I miss the focal blur. Comments? GDallimore (Talk) 18:52, 5 January 2011 (UTC)[reply]
        Version 2 looks good. I think it is better without the focal blur as it makes the edge of the sphere sharper, thereby emphasizing the differences in the two rendering techniques. ~~However, the shadows in Version 2 look wonky (looks like there are many point light sources).~~ Purpy Pupple (talk) 00:41, 6 January 2011 (UTC)[reply]
        Thanks. In response to your now deleted comment, the area lighting setting were quite low but that is easily fixed. GDallimore (Talk) 20:49, 13 January 2011 (UTC)[reply]
        Final version of "Version 2" uploaded with slightly improved area lighting, but otherwise essentially identical. GDallimore (Talk) 21:05, 13 January 2011 (UTC)[reply]
Comment. Why are there no chocolate chips visible? Kaldari (talk) 21:17, 4 January 2011 (UTC)[reply]
- Maybe because the picture depicts this and not a chocolate chip cookie? lol. Purpy Pupple (talk) 22:02, 4 January 2011 (UTC)[reply]
  - Yes, I was worried it was slightly dung like and actually fiddled with the shades of brown for ages to try to steer away from that! On another light note, I almost made the floor checkered in honour of raytracing cliche of reflective spheres over checkered planes, but it distracted from the shadows. GDallimore (Talk) 23:02, 4 January 2011 (UTC)[reply]

Comment. I can't tell what the difference is between the two balls. The caption seems to be discussing a difference but then the images look alike. Also, although I am not a mathematician, I think I should probably be able to understand most technical explanations if an effort is made to make them accessible. But this one baffled me. Consider writing the caption to make it discuss the features of the images (or make the images look more different). And write the caption so that it is not so dependent on out of article link clicking. Maybe it is impossible, but give it a shot. Can you convey the "so what", more clearly?TCO (talk) 09:22, 12 January 2011 (UTC)[reply]

I reread it and understand the concept now. One way has a perfect circle, the other doesn't. Consider making it simpler or quicker to understand though. The other thing is part of the caption discusses "changing the actual surface" in a manner that suggest physical actions. But the whole thing is a model. So this wording could be improved (common issue with modelers).TCO (talk) 09:28, 12 January 2011 (UTC)[reply]

I'm not pushing this, but maybe if you had some red dashed line or the like for the outline, it would help make it clear faster for blockheads like me, what the key point is. Other than that thee bumps look great and so the shadows.TCO (talk) 09:31, 12 January 2011 (UTC)[reply]

I agree it's hard to spot some of the key differences in the thumbnail, but I don't agree that it's difficult in the full-size or even preview versions. The differences are always going to be quite subtle, since bump-mapping is a pretty good technique despite its limitations, so I don't think there's any way to really make the differences stand out in the thumb version without really pushing the technique to its limits so that the bump-mapped sphere just looked terrible. I don't think the lines and circles you suggest adding would help. I'll have a final think about the blurb - I tried to be pretty careful about the whole "modelling" thing, but see a couple of places it can be improved, so thanks. GDallimore (Talk) 20:49, 13 January 2011 (UTC)[reply]

Friend, just fix the blurb and it is copacetic (and I will support you). I agree the red dashed lines would be bafugly. And yeah, I can easily see "it" when I know what to look for. But weeding through all that spinach about the programs and all makes it hard to get the key aha right away. Look how simply I expressed it when I finally understood it was circle verus non-circle. I would not write a caption like that even in a formal science paper, nor would I do it in Wikipedia (although oddly sometimes Wiki acts more technical than real technical articles). I remember reading something on "how to write science papers" (some book or such) that figure captions are some of the most high value text you have in your entire manuscript, so hone them. I think you can actually keep all the detail, but just say something at the very front to the effect of "Bal on the left has a perfectly circular outline while the one on the right has a bumped outline." (then still include all the stuff about the programs and all in later sentences). But call the reader attention to the key visual aspect of the picture first, rather than making him read through all that stuff, think, look, think, look. Also, I think referring to the surface as if it were an actual physical object is confusing (and wrong, there was a great review paper done on how mathematical modeller's language of discussing things as reality can actually affect their work negatively.) Anyhow. Good luck, man. It's really no big deal, just sharing in case it helps.TCO (talk) 20:59, 13 January 2011 (UTC)[reply]

Sorry if my reply came across as suggesting you were being unhelpful. That really wasn't the intention and my "thanks" were meant honestly. And the suggestion to get the main point across early is a great one. GDallimore (Talk) 21:05, 13 January 2011 (UTC)[reply]

Weak support I've thought quite a bit about this image the past couple days. I really like the educational value of the image; it definitely adds a huge understanding of the concept to a layman. That said, it's hard for me to fully support the image for FP, as it doesn't seem particularly "eye-catching". I've tried to think of a way to really make the image grab you a bit more, and let the viewer understand the concept without close inspection, but I haven't come up with anything. Great work though. Jujutacular ^talk 20:21, 13 January 2011 (UTC)[reply]
- Yeah, that's why I tried the fog in the original version. The problem is, do anything more to make it more eye-catching and you lose the point of the image. Amusingly, with the plain white background, it now looks like one of those dissected fruit photos! GDallimore (Talk) 20:49, 13 January 2011 (UTC)[reply]
  - Of course, the original red colouring I used is a bit more eye-catching, but I though it was too much and went for muddy realism. What do you think of this? GDallimore (Talk) 20:57, 13 January 2011 (UTC)[reply]

Not Promoted --Makeemlighter (talk) 20:54, 13 January 2011 (UTC)[reply]

Not enough support. Makeemlighter (talk) 20:54, 13 January 2011 (UTC)[reply]