AuthorTopic: Second Cluster Study - Knights of the Round  (Read 36296 times)

Offline Helm

  • Moderator
  • 0110
  • *
  • Posts: 5159
  • Karma: +0/-0
    • View Profile
    • Asides-Bsides

Second Cluster Study - Knights of the Round

on: January 01, 2013, 05:55:01 pm
Hello, hello!

I've talked a lot about clusters and all my theories and examples involve the new school clean pixel displays we now work on. However pixel art came to be mostly on old bleedy CRT monitors both for computers and  - more importantly for us - on arcade cabinets. Many of these CRTs running all day were blown out all to hell. Phosphorous glow would leave red trails after sprites, curvy screens would deform the art, flickery scanlines would change pixel values, and non-standard pixel ratios would squish or stretch accordingly.

We won't be emulating all of these effects, but even at the best of times, The artist would have to live with phosphorous glow, scanlines and different aspect ratios.

What I want you to do is take a look at  Capcom arcade classic, Knights of the Round.

We'll be focusing on the Arthur sprite in particular



It's interesting, isn't it? Running on CPS1 hardware. Like all (most?) old arcade cabinets, it has a weird pixel aspect ratio. 12:7, 384x224 pixels,  but still stretched into 4:3 screen aspect (something CRTs are a lot better at than modern non-CRT ones). It has a 16 bit color range, which is pretty much limitless for our intents. The original sprite has 26 colors. CPS1 was really powerful, compared to any 16 bit home console! I was wrong, Carnivac below catches the the sprite is 16 colours without the sword! So feel free to work with a 16 color palette just for the sprite either with a weapon separate (good time to make your own completely original weapon!) and its separate weapon palette, or one using the same palette as the sprite for extra challenge!

What I want you to do is to edit this to your heart's content. Change the colors and use whatever shading method you'd like. Just try to keep it within the aesthetic context of the old arcade game as you understand it. For my edit I based my changes on the thought that this game is trying to convey the bright, well-light context of a board game and the characters are pawns on a meticulously crafted battle map. My theory is supported by the in-between level assessment with people shuffling figures about. I took it as an approximation of a paper and pencil and lead figurine dungeons and dragons-esque session. So I tried to color and tint my sprite like someone would paint a lead figurine. Keeping the strong, saturated primaries but adding tints and dirt and stuff. Still not making it more realistic, just adding more clarity and refining the areas of complexity at the same time. I removed a lot of colors, as well.

You can look at my edit after you've done yours by clicking here.
And then you can look at the filtered version. We'll be filtering your edits Pixelation-side as well. Don't bother trying to filter them on your own, the exercise is exactly about not knowing how your art would look on an old CRT first, and then comparing and picking up practical knowledge. I have many notes on what I've learned from comparing the filtered version to the crisp edit, but I'll save them for after we've done a round of edits.

HUGEASS scanline filtered image!
NEW AND BETTER SCANLINES
« Last Edit: January 05, 2013, 01:51:41 pm by ptoing »

Offline st0ven

  • 0010
  • *
  • Posts: 200
  • Karma: +0/-0
    • View Profile
    • spriteart

Re: Second Cluster Study - Knights of the Round

Reply #1 on: January 01, 2013, 07:16:00 pm
interesting analysis. Ill be excited to observe your edit and also comparisons on post-process filtering.

After examining the original sprite simply from a palette perspective, im already noticing that one of the first things ill be seeking to do is adjusting values to separate the gloves from skintone. If you are running on hardware that can handle at least an 8 bit sprite, it seems to gain very little by needlessly sharing such values so heavily, particularly when the darkest value offers inadequate contrast to differentiate proper depth and distinction.

One thing ive learned that i really dislike about capcoms 90's games in their spritework was the need to 'shift' hues (particularly skintones) that would suggest a magenta-like atmospheric ambiance. Ive become much more fond of skewing towards a more natural blue ambiance, 'shifting' hues essentially in the other direction. Ill be adjusting my palette immediately to reflect that.

Offline Carnivac

  • 0010
  • *
  • Posts: 269
  • Karma: +0/-0
  • Mayhem Attack Squad
    • View Profile
    • Doctor Who - Retro Sprite Gallery

Re: Second Cluster Study - Knights of the Round

Reply #2 on: January 01, 2013, 07:35:37 pm
I always assumed weapons in these sorts of games such as that sword he's holding, was a seperate sprite with it's own palette.  Maybe I was wrong then.

I just checked by removing the sword and the knight alone is 16 colors which seemed right.  If I was to do this I would probably do the same.  16 colors for the guy but then I'm trying to keep most of my sprites 16 colors or less to get better at palette usage.
« Last Edit: January 01, 2013, 07:37:59 pm by Carnivac »
NES, Amiga & Amstrad CPC inspired
I know nothing about pixel art
http://carnivac.tumblr.com/

Offline ptoing

  • 0101
  • ****
  • Posts: 3063
  • Karma: +0/-0
  • variegated quadrangle arranger
    • the_ptoing
    • http://pixeljoint.com/p/2191.htm
    • View Profile
    • Perpetually inactive website

Re: Second Cluster Study - Knights of the Round

Reply #3 on: January 01, 2013, 07:55:59 pm
Carnivac might be right. I think the CPS1 can not do more than 16 colours per sprite.

Edit: Checked. Yeah. 256 sprites total, they are apperantly 16x16 and can have 15+trans.
« Last Edit: January 01, 2013, 07:58:38 pm by ptoing »
There are no ugly colours, only ugly combinations of colours.

Offline st0ven

  • 0010
  • *
  • Posts: 200
  • Karma: +0/-0
    • View Profile
    • spriteart

Re: Second Cluster Study - Knights of the Round

Reply #4 on: January 01, 2013, 08:02:23 pm
that does make a lot more sense. CPS1 was a relatively older board compared to the newer fighting/action games produced by capcom.

in that case - it would also stand to reason that there is probably little value in editing this sprite as a whole considering the tech involved in combining the two sprites to be one image - where palette is undoubtedly an important aspect of the exercise. Perhaps create a new template to work with from which the sprite and weapon are separated? or at least set some base level of common understanding that the weapon palette must maintain unique from the sprite palette and the sprite should not go above 15 non-transparent colors (same for the weapon).

Good catch Carnivac.

Offline Lóng

  • 0001
  • *
  • Posts: 40
  • Karma: +0/-2
  • Sometimes the scary spider is just a pile of dust.
    • View Profile

Re: Second Cluster Study - Knights of the Round

Reply #5 on: January 01, 2013, 08:22:07 pm
Argh and here I just finished my edit, reduced the overall color count to 13, raised the contrast and  tightened up some clusters, while trying to retain the original look aside from making the face more menacing.

so if I change the sword-palette to a completely new one, that has less than 15 colors it should be ok?
« Last Edit: January 02, 2013, 08:35:20 am by Lóng »

Offline 0xDB

  • 0011
  • **
  • Posts: 873
  • Karma: +0/-0
  • Dennis inter-is.
    • dennisbusch_de
    • http://pixeljoint.com/p/1287.htm
    • 0xdb
    • View Profile
    • 0xDB

Re: Second Cluster Study - Knights of the Round

Reply #6 on: January 01, 2013, 09:38:42 pm
(Disclaimer: Haven't been following or studying the pixel cluster theory development so my edit probably does not follow the theory at all. Also haven't pixeled anything for a while but this was fun and a welcome distraction from drawing naked people.)

Kept all the original colors as they were. My goal was to improve upon the volumes and the textures of the original:

Offline Carnivac

  • 0010
  • *
  • Posts: 269
  • Karma: +0/-0
  • Mayhem Attack Squad
    • View Profile
    • Doctor Who - Retro Sprite Gallery

Re: Second Cluster Study - Knights of the Round

Reply #7 on: January 01, 2013, 10:59:49 pm
Carnivac might be right. I think the CPS1 can not do more than 16 colours per sprite.

I actually got something right?  Am amazed.   Was oddly coincidental because for last few days I've been designing some sprites that are 16 color and then have a weapon sprite overlayed with it's own palette (I say it's own, but it can reuse colors already present in the image but point is it's a seperate sprite from the warrior and niether exceeds 16 colors).  I just thought old arcade games did the same partially for palette reasons but because it's more practical to have a weapon as a seperate sprite especially if there are a selection of different weapons held the same way.
NES, Amiga & Amstrad CPC inspired
I know nothing about pixel art
http://carnivac.tumblr.com/

Offline Helm

  • Moderator
  • 0110
  • *
  • Posts: 5159
  • Karma: +0/-0
    • View Profile
    • Asides-Bsides

Re: Second Cluster Study - Knights of the Round

Reply #8 on: January 02, 2013, 12:21:01 am
Good catch! I have amended the original write-up with new guidelines. I will also edit my edit though I'm fairly certain the sprite I've done on its own isn't hard to get to 16 colors.

Offline Cure

  • 0011
  • **
  • Posts: 565
  • Karma: +0/-0
    • http://pixeljoint.com/p/2621.htm
    • facebook.com/logantannerart
    • View Profile

Re: Second Cluster Study - Knights of the Round

Reply #9 on: January 02, 2013, 02:46:40 am
edit:
« Last Edit: January 05, 2013, 11:55:38 pm by Cure »