1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Eirin Yagokoro from Touhou

Discussion in 'SDT Requests' started by Klue, Sep 17, 2015.

  1. Klue

    Klue Potential Patron

    Joined:
    Oct 13, 2014
    Messages:
    10
    Likes Received:
    1
    Hello! I would like to request for Eirin Yagokoro from Touhou

    Would anyone be interested in making her?

    Imgur album for side refs of her http://imgur.com/a/fDUlD
     
  2. stuntcock

    stuntcock Club Regular Staff Member Moderator

    Joined:
    Jun 5, 2012
    Messages:
    1,524
    Likes Received:
    356
    You'll need better references; the images that you've chosen show too much inconsistency. The hair colors include various blends of white, grey, green, and purple. The braiding varies considerably in style and thickness (one image even shows loose hair with no braiding at all!). Different hats are worn. The girl's ear is visible in some shots, but in others it is fully covered by hair. The bangs flip between concave and convex.

    Even if we ignore the inconsistencies, most of your images don't fit the SDT art style. They use flattened colors, excessively soft lighting, heavy backlighting, or inappropriate light sources. One image uses the wrong camera angle and another is much too small.

    Two of your images could be used as references. But the first includes too much visual detail (it would take ages to trace). The second has a reasonable amount of visual detail, but most of the actual hair has been removed by cropping or occulted by the girl's body.



    Instead of offering a half-dozen different incarnations of the character, try to find one really good Reference Image and stick with it. Include additional images only to clarify minor details which are absent or poorly-detailed in the main image (such as the hat or the hair ribbons).
     
  3. Klue

    Klue Potential Patron

    Joined:
    Oct 13, 2014
    Messages:
    10
    Likes Received:
    1
    Oh, I see. Then would the second reference image you mentioned be suffice and for the hair and the hat, I could find other images for them? Since the hair is cut from the image and the hat is inconsistent

    About her hair, I read here that she has silver hair so I think her color leans toward white than grey
     
  4. stuntcock

    stuntcock Club Regular Staff Member Moderator

    Joined:
    Jun 5, 2012
    Messages:
    1,524
    Likes Received:
    356
    The second image seems useless to me, because most of the hair is hidden and it's the wrong color anyways (blue-grey instead of silver).

    I suppose that we could use it as a reference for the charcode (skin color, eye color, breast size, etc). But those parameters are very easy to customize; we usually don't need special reference images for them.

    If you want the request to proceed then you'll need a side-view reference image which shows the full hairstyle. Whether or not this image includes a hat is at your discretion. If it does include a hat, then the hat would probably be included as a permanent (non-removable) feature of the hairstyle. It the hat is missing from the reference image, then it would presumably become a separate mod (based on an additional reference image) - users would be able to toggle the hat on and off.



    Edit:

    That's a sketch. It has low resolution and some details are simplified (such as the wrap/ribbon at the end of the braid). If we look only at the lines, then it gives us a complete shape for the hair, as well as its proportions w/r/t the head and body.

    Unfortunately, the brushstrokes aren't really usable as-is. They're approximate - they tend to stray outside the lines. There are also many gaps, where the brushstroke does not reach the actual outline. Some of these gaps seem intentional (leaving whitespace to indicate bright/reflective edges) but other gaps seem to be stylistic or simply lazy.

    The bottom line is that an SDT artist would need to review the entire thing and apply artistic judgment. They'd need to clean up the boundaries between highlight and main-fill (redrawing some of the highlights so that they point towards a consistent light source), and then add shadow details (because the reference image doesn't include any shadows at all).



    An actual artist (like Zoltan, Mineur, Dante, et al) could probably create a hairstyle based on the images that you've provided. Those guys tend to be busy. If you're willing to wait a few weeks, then one of them might take an interest in your request.

    If you want faster results, then you're going to need to find a reference image which can be traced without the need for artistic skill or interpretation. For comparison, here is an image that I traced recently. Note that it's a high-resolution image; it uses three colors (highlight-midtone-shadow); the lines and fills are fully aligned; the overall lighting pattern is similar to those of existing SDT hairstyles. Hence, it's something that can easily be traced and made into a hairstyle for SDT, even by someone who lacks artistic skill.
     
  5. Klue

    Klue Potential Patron

    Joined:
    Oct 13, 2014
    Messages:
    10
    Likes Received:
    1
    Hmm, how's this?

    Edit:

    Does the girl need to be facing right side? And how much backlighting is too much? Because if not, I saw this reference.

    Probably the only issues with the reference is that not all her hairstyle is shown (although most of it is shown) and (if it's too much) the backlighting. So I dunno if that's a good enough reference
     
  6. Klue

    Klue Potential Patron

    Joined:
    Oct 13, 2014
    Messages:
    10
    Likes Received:
    1
    Sorry double post
     
  7. stuntcock

    stuntcock Club Regular Staff Member Moderator

    Joined:
    Jun 5, 2012
    Messages:
    1,524
    Likes Received:
    356
    Does it matter? Yes, but only insofar as the actual hairstyle is asymmetric, and insofar as you care about presenting the "correct" side to the viewer.

    The original character design seems to show a center-parted symmetric hairstyle. The hair shown in the "classroom" reference image is noticeably asymmetric (the background side of her bangs is much thicker than the foreground side), but this doesn't present any technical challenges. If you can accept the design being imported to SDT as-is, then we can proceed with it.

    This image is acceptable. It has a proper palette, with clearly-defined areas of fill, highlight, and shadow -- and its highlights indicate a single, consistent light source. Backlighting is troublesome only when it creates excessive highlights (e.g. a "halo effect" around every edge), or when it causes fill boundaries to bleed.

    My only concern with this image is that the scene lighting is somewhat diffuse and warm, suggesting sunset illumination. This may mean that the colors shown are not 100% accurate to the character design. This is not a serious problem, because it will take a while to trace the vectors. During that time, you can fiddle with graphics programs (such as Photoshop or GIMP) if you'd like to to adjust the color balance. Here's a quick example of desaturation and rebalancing, aiming to reduce the purple tint of the hair.

    Note: raster graphics operations will tend to degrade the image quality. The image that I linked above is worse than the original. That's ok! The actual vector tracing work will rely on the original image. The modified version (if you choose to provide one) will be used only when assigning fill colors to the completed vectors. So there's no harm if the modified image becomes pixellated or blurry.

    Note: image manipulation is not necessary. If you're happy with the purple tint, then we can leave it as-is.

    This will work. A few caveats:
    • when dealing with reference images, please link to the highest-resolution version that's available. Vector tracing is often done at high zoom levels; better resolution means fewer difficulties with aliased lines.
      • if you're uncertain about which version is best, or you don't know whether a direct imagepath link is allowed, then linking to the source page (or the imageboard post) is the best option.
    • the hat isn't shown in side view; we can see the entirety of the red-cross badge, instead of seeing only half of it.
      • we could just trace the hat as-is. Assume that she's wearing it at a funny angle.
      • we could trace the hat as-is, but shift the red cross to the front and cut it in half.
      • we could omit the hat entirely while tracing, and then replace it with a hat from a different reference image.
        • Note: this is based on the assumption that the hat is bad/wrong, because it lacks the solid front and the "constellation" lines shown in several other images. If you're happy with the hat shown in this image, then there's no need to replace it.
    • the hairstyle is fairly large and detailed; it will take me a while to trace.
    • the reference image doesn't show a neutral standing or kneeling posture; the girl is leaning forward and her braid is resting against her back. I'll need to straighten the braid while rigging it.
      • the first step is to trace the image as-is; any manipulation of its shape will occur later.
      • I'll post WIP screenshots at that point, showing the length and curvature of the braid w/r/t the kneeling SDT template. You'll have an opportunity to make suggestions or offer feedback.
    • I won't make any promises about animation.
      • by the time that I'm finished tracing and rigging this thing, I may be completely sick of looking at it.
      • the physics properties of the bangs and sidelocks can be inferred from their static position, and from their length.
      • since we have no animation references, the braid would require a lot of assumptions (w/r/t weight, tightness of braiding, hair elasticity, cross-sectional density, etc).
        • I found some MMD scenes which include this character, but the hair rigging was pretty poor. If I decide to animate the hairstyle, then I may ask you to track down a few decent video clips.
      • it would require extra work to make the braid lie flat against the girl's back when she leans forward, instead of clipping through her torso.
     
  8. Klue

    Klue Potential Patron

    Joined:
    Oct 13, 2014
    Messages:
    10
    Likes Received:
    1
    Yup! I accept the design as is


    Hmm, then if you're ok with it, can you use the desaturated edit of the reference image you made? [/quote] because I think your edit looks excellent already




    • Alright. I'll keep that in mind in case I make another request

      I see. I think tracing the hat as-is but shifting the red cross to the front is fine. Let's go with that

      No problem! ;D

      Alright. I'll look at the first output once it's done

      By this, do you mean the hair would be swaying and moving? Because I would be fine with a static version :)
     
  9. stuntcock

    stuntcock Club Regular Staff Member Moderator

    Joined:
    Jun 5, 2012
    Messages:
    1,524
    Likes Received:
    356
    [​IMG]

    So. Many. Fucking. Lines.



    Yes. I've half-assed the vector work - for example, there aren't enough layers to allow independent movement of the twelve strands which compose the sidelocks. This was done because it isn't feasible to fully animate the hairstyle -- even if I was willing to accept the workload, the resulting segment count (60+) would incur an unacceptable impact on the game's performance (FPS).

    We have enough layers for partial animation (e.g. braid, bangs, sidelocks ... perhaps even a slightly-bouncy hat). But I'd feel weird about doing animation work on a project that I don't intend to actually finish ... so I'll probably just merge the layers and then publish it as a static SWF file.

    The shape of the hat is guesswork, because the reference image is cropped off. I normally would have filled in the gap by checking the other references, but they're completely inconsistent w/r/t the size and shape of the hat (let alone color and shading details). If you can find a reasonable reference then post it; otherwise we're stuck with my extrapolation.

    I need to redraw the ear-to-nape section of the hair, because the SDT skull is somewhat misshapen.

    The over/under layering around the ear has been preserved, although the SDT ear is proportionally larger than the one shown in the reference image.



    Edit: I give up.

    The in-game version of the hair looks significantly worse than what I had anticipated. It doesn't seem worth the effort to finish tracing the braid (or even tackling the minor tasks, such as "optimize the fills and layers"). Especially since I've belatedly noticed that this thread has zero "Like"s and so I'm doing hours of work for one person (which is a terribly inefficient ratio, even if our one consumer is the nicest guy in the world). And the project doesn't even include any animation work, which is the stuff that I actually enjoy.

    If anyone wants to take over then here's the FLA.
     
  10. Klue

    Klue Potential Patron

    Joined:
    Oct 13, 2014
    Messages:
    10
    Likes Received:
    1
    I think it already looks fine. Like you said, the images show inconsistency so I think it already fits

    Thanks for taking up my request btw!