AuthorTopic: fruity shading experiments  (Read 17056 times)

Offline Ai

  • 0100
  • ***
  • Posts: 1057
  • Karma: +2/-0
  • finti
    • http://pixeljoint.com/pixels/profile.asp?id=1996
    • finticemo
    • View Profile

Re: fruity shading experiments

Reply #10 on: July 01, 2012, 10:03:12 am

A core shadow? I've never heard of this concept. Is it a shadow that, in conjunction with the outline,defines the 3d shape of the object?

Yeah, that's it.  I used to have a very nice picture of an egg with the shadows and whatnot labeled, but I can't seem to find it at the moment.  So here's the same info in a long tutorial:



with the most relevant image here (linked for size):

http://www.conceptart.org/wiki/lib/exe/fetch.php?w=&h=&cache=cache&media=sphere_labeled.jpg

Human vision works on contrast.  If you smooth away all the contrast, the result looks a bit smudgy.
So, are you suggesting that the cumquat and plum also suffer from this problem? or just the lemon?
Personally I feel that if I substantially kill contrast, I've performed the method wrong.. it's supposed to detail the shading, not make everything smooth per se. (after all, that example image is full of continuous gradients itself, but it apparently has a terminator/core shadow.)
I'll try a few shading variations and post them.. previously, I'd had a darkening on the top left edge, which I discarded because it seemed to be just looking weird instead of improving 3dness.

Quote
Maybe try a fractal approach?  Where you have smooth shading within each light/medium/dark area, but still keep a larger jump between the different areas.  That might give you a vexel style look though, which isn't what you're going for.  Dunno, jut thinking off the cuff. 
IMO on the lemon at least, the problem is that the subdivisions of light level are too evenly spaced, they need variation.. AFAICS a zillion subdivisions are okay and shouldn't significantly effect contrast in important areas, -when properly performed-.. to clarify, the subdivisions usually should not be even.. The cherries are a good example of that. I'm getting the impression that I should totally rerender the lemon.

Quote
I like the experiment and hope to see it continue.
Tourist
Thanks; yeah, I'll be able to continue working on it and posting it.. my computer is fine, just, physical access to it is currently intermittent.


Quote from: Bissle
Whenever you get more time, could you take a pic of the plums in your backyard? They sound interesting. Looking forward to more progress on this
No chance of that  -- it's not plum season and the house is going to be sold by the time it is plum season.
However.. I have photographed them in the past, and archive all my photos, so I should be able to dig some up.
« Last Edit: July 01, 2012, 12:21:35 pm by Ai »
If you insist on being pessimistic about your own abilities, consider also being pessimistic about the accuracy of that pessimistic judgement.

Offline Tourist

  • 0010
  • *
  • Posts: 376
  • Karma: +1/-0
    • View Profile

Re: fruity shading experiments

Reply #11 on: July 01, 2012, 02:53:29 pm
Quote
So, are you suggesting that the cumquat and plum also suffer from this problem? or just the lemon?

The kumquat does not suffer from this because the light is almost dead center.  There's not much shadow that will show up.

The plum and cherry do not suffer from this because there is a strong contrast between light and dark areas, though to be honest they both kind of fade into the background.  That could be my monitor though.  But yeah, they look fine.

Also, edited the tutorial link in my previous post, should work now.

Quote
IMO on the lemon at least, the problem is that the subdivisions of light level are too evenly spaced, they need variation..
I agree with this.  Here is a photo of a lemon with similar lighting to your image.  It has a distinct separation of light and shadow, like the labeled image I linked to earlier.  Your latest version has some of this, but I think it should be stronger.  Knock down the values on the shadow side of the lemon by a step and it will look better, imo.



About the process, do you think it would work equally well for palettes using LAB that incorporate more uniform steps in lightness?  Or does the linear RGB scale (non-linear lightness) contribute to the effect?

Tourist

Offline Ai

  • 0100
  • ***
  • Posts: 1057
  • Karma: +2/-0
  • finti
    • http://pixeljoint.com/pixels/profile.asp?id=1996
    • finticemo
    • View Profile

Re: fruity shading experiments

Reply #12 on: July 02, 2012, 01:46:48 pm
Quote
So, are you suggesting that the cumquat and plum also suffer from this problem? or just the lemon?

The kumquat does not suffer from this because the light is almost dead center.  There's not much shadow that will show up.

The plum and cherry do not suffer from this because there is a strong contrast between light and dark areas, though to be honest they both kind of fade into the background.  That could be my monitor though.  But yeah, they look fine.
Thanks, this is useful feedback telling me I need to improve my palettes (maybe add a second light source or increase hue shift). I tried to only shade naturally and not intentionally AA into the bg.

Quote
Quote
IMO on the lemon at least, the problem is that the subdivisions of light level are too evenly spaced, they need variation..
I agree with this.  Here is a photo of a lemon with similar lighting to your image.  It has a distinct separation of light and shadow, like the labeled image I linked to earlier.  Your latest version has some of this, but I think it should be stronger.  Knock down the values on the shadow side of the lemon by a step and it will look better, imo.



Excellent example, thanks!
Quote

About the process, do you think it would work equally well for palettes using LAB that incorporate more uniform steps in lightness?  Or does the linear RGB scale (non-linear lightness) contribute to the effect?

Tourist

LAB is near ideal (accurate perceptual inbetweens are best IME). I happen to be using RGB gradienting because I've chosen to use a restricted palette colorspace (64 intensity levels), and also cause I didn't want to bother with generating the new shades via GPick, just work on my basic technique.

EDIT: My previous work on Bub-n-Bros used LAB color mixing, now that I think of it, and while I think my technique could have been improved, I was happy with the overall color ramps created. The art is available in PPM format as part of the game download from http://bub-n-bros.sf.net ; one example of the method is the bubbles themselves -- 100 distinct shades used in a .. 32x32 sprite, I think? The bonuses also used the technique heavily. I'll make an edit with some examples if it seems I have the right software available on this laptop.
« Last Edit: July 03, 2012, 12:47:42 am by Ai »
If you insist on being pessimistic about your own abilities, consider also being pessimistic about the accuracy of that pessimistic judgement.

Offline Grimsane

  • 0010
  • *
  • Posts: 423
  • Karma: +2/-0
    • View Profile

Re: fruity shading experiments

Reply #13 on: July 03, 2012, 04:23:10 am
I think that approach on the colouring of the lemon just feeds into the generally unpleasant and murkiness that is dark pure yellow

something like this is both more realistic and more appealing on the eye in my opinion


which brings me to another piece of critique, I think the fruit looks a bit flattened and warped in your sketches and consequently your pixels, some more than others, and even if the reference was some rather oblong fruit it should be avoided because it's not the widely associated traits of said fruit, I've shown one suggestion on reintroducing more a more round and volumetric form in the edit.

the closest thing I could think of to explaining what I'm seeing is this:



in this example, imagine you are trying to achieve a sphere and you let people know you are trying to convey a sphere, if it wasn't known by other's what you are trying to achieve it wouldn't look off, but once we have a point of reference and know what it is intended to be they realize it doesn't look quite right, knowing this is what you intended



also there is also a case of not terminating the edge of the object's silhouette properly and stretching details out, which make the viewing angle look quite off, and it's what alot of artists go through in my general observation, i struggled with it early on too, here's something to illustrate



hope it conveys what I'm observing, just making you aware of it, because it'll be all but impossible to address if you weren't aware of it.

other than that the colour and texture looks pretty solid continue  :y:

Offline Ai

  • 0100
  • ***
  • Posts: 1057
  • Karma: +2/-0
  • finti
    • http://pixeljoint.com/pixels/profile.asp?id=1996
    • finticemo
    • View Profile

Re: fruity shading experiments

Reply #14 on: July 03, 2012, 06:37:19 am
I think that approach on the colouring of the lemon just feeds into the generally unpleasant and murkiness that is dark pure yellow

something like this is both more realistic and more appealing on the eye in my opinion



which brings me to another piece of critique, I think the fruit looks a bit flattened and warped in your sketches and consequently your pixels, some more than others, and even if the reference was some rather oblong fruit it should be avoided because it's not the widely associated traits of said fruit, I've shown one suggestion on reintroducing more a more round and volumetric form in the edit.

I think what is correct must be something in between. Given the many real life references I have here, my version looks skinny and yours looks chubby ( I have also really fat lemons occasionally here.. they scarcely look different from grapefruit. Obviously they fail the iconicness test.). Similarly, I intentionally gave a green tinge to the lemon's darker shades, but as you point out this doesn't work, OTOH your edit takes the goldness too far (for iconicness) and makes it look like a washed out orange*.  The bulge on the end of yours is spot-on though, definitely my 'point' was too pointy.

I'll probably end up making at least 2 ramps -- a semi-goldish one and a semi-greenish one, and mix them 80/20 across the lemon's surface.
Since I have a largish set of citrus on my list, you'll be able to judge for yourself how well I separate their characteristics iconically.

* FWIW I'm currently forced to view this through a laptop screen -- far from ideal :)
Also, 'dark pure yellow' is actually an artefact of the nonlinearity of sRGB. if you make a gradient yellow->black using LAB (eg. with the help of GPick), the colors produced are considerably more pleasing than the corresponding gradient in sRGB, which is actually unreasonably greenish.

Quote
the closest thing I could think of to explaining what I'm seeing is this:



in this example, imagine you are trying to achieve a sphere and you let people know you are trying to convey a sphere, if it wasn't known by other's what you are trying to achieve it wouldn't look off, but once we have a point of reference and know what it is intended to be they realize it doesn't look quite right, knowing this is what you intended
I take your point, regardless of my view on the exact details of your specific edit.

Quote


also there is also a case of not terminating the edge of the object's silhouette properly and stretching details out, which make the viewing angle look quite off, and it's what alot of artists go through in my general observation, i struggled with it early on too, here's something to illustrate



hope it conveys what I'm observing, just making you aware of it, because it'll be all but impossible to address if you weren't aware of it.
Yes, I can see this is very true! I'll definitely take that to heart, and work on defining the edges more truthfully. Very much appreciated :)
« Last Edit: July 03, 2012, 07:05:58 am by Ai »
If you insist on being pessimistic about your own abilities, consider also being pessimistic about the accuracy of that pessimistic judgement.

Offline Grimsane

  • 0010
  • *
  • Posts: 423
  • Karma: +2/-0
    • View Profile

Re: fruity shading experiments

Reply #15 on: July 03, 2012, 07:11:39 am
yeah it wasn't a definite suggestion on the size or proportion to take your lemon, just an exaggeration to emphasis and point out that from my observation it seemed a tad too elongated, relative to any lemon I've ever seen, and the reference posted, and yeah I just used the curve editor and jacked up the low end of the red channel rather swiftly again another exaggeration to make my point obvious,

regarding those elaborations both were the same point really I just realized the first one was really poorly illustrated so I made another  :) glad my observations weren't entirely off and have been taken constructively, made the observation when you first posted this, but was hesitant to bring it up because it's some pretty critical criticism, but it is good to have things like that pointed out instead of ignored all in the name of improvement

oh and btw the most common lemons here look like that edit :lol:

Offline Ai

  • 0100
  • ***
  • Posts: 1057
  • Karma: +2/-0
  • finti
    • http://pixeljoint.com/pixels/profile.asp?id=1996
    • finticemo
    • View Profile

Re: fruity shading experiments

Reply #16 on: July 03, 2012, 11:55:33 am
yeah it wasn't a definite suggestion on the size or proportion to take your lemon, just an exaggeration to emphasis and point out that from my observation it seemed a tad too elongated, relative to any lemon I've ever seen, and the reference posted, and yeah I just used the curve editor and jacked up the low end of the red channel rather swiftly again another exaggeration to make my point obvious,
Good to know.

Quote
regarding those elaborations both were the same point really I just realized the first one was really poorly illustrated so I made another  :) glad my observations weren't entirely off and have been taken constructively, made the observation when you first posted this, but was hesitant to bring it up because it's some pretty critical criticism, but it is good to have things like that pointed out instead of ignored all in the name of improvement

oh and btw the most common lemons here look like that edit :lol:
Commercial lemons are always super gold (as lemons go) and plump, so no surprise. What I see in people's backyards hasn't matched that image even once so far :) Anyway I'll do a bush lemon (image link) as well, which will be fascinating to render.. they're commonly that gold, so we can both be happy  :lol: (also, finding images of them online is really hard, wow)
If you insist on being pessimistic about your own abilities, consider also being pessimistic about the accuracy of that pessimistic judgement.

Offline Ai

  • 0100
  • ***
  • Posts: 1057
  • Karma: +2/-0
  • finti
    • http://pixeljoint.com/pixels/profile.asp?id=1996
    • finticemo
    • View Profile

Re: fruity shading experiments

Reply #17 on: July 04, 2012, 02:24:46 pm
Old:

New:

Fah, this transition to a new setup has been strange. I seem to be flailing away at this far more than purposefully doing stuff. I think I misdistributed the shades on the left/upper side. Well, I've been working at this awhile now, hope it's better in at least shaping :) Any other C+C welcome too.

Edit: newest:


Edit3:
Trundle trundle.
It's getting there..

That segment at the top is giving some serious trouble. Any suggestions?
« Last Edit: July 05, 2012, 09:12:16 am by Ai »
If you insist on being pessimistic about your own abilities, consider also being pessimistic about the accuracy of that pessimistic judgement.

Offline Tourist

  • 0010
  • *
  • Posts: 376
  • Karma: +1/-0
    • View Profile

Re: fruity shading experiments

Reply #18 on: July 07, 2012, 04:53:48 am
Maybe use the sawtooth / square wave style dither to break up the banding at the top?  Definitely looking better though.

Tourist

Offline Bissle

  • 0010
  • *
  • Posts: 117
  • Karma: +0/-0
    • View Profile

Re: fruity shading experiments

Reply #19 on: July 07, 2012, 10:24:27 am
The lemon is looking supermarket fresh (whatever that means)! Like Tourist said, maybe dithering the lemon in that style would help with the banding.