It is currently Sat Jul 06, 2024 7:53 pm


All times are UTC - 5 hours [ DST ]



Post new topic Reply to topic  [ 8 posts ] 
Author Message
 Post subject: Do you need vector graphics to get better sphere illusions?
PostPosted: Wed Apr 02, 2014 7:06 am  (#1) 
Offline
GimpChat Member

Joined: Mar 24, 2014
Posts: 21
I experimented with bump-mapping of a disk, and with some trial and error I get a fairly convincing dome (I'd show it but the system is giving me some weird flak over posting it even as a png, after I forgot to convert it from xcf the first time, and still after I renamed it - you've probably seen this effect before anyway). I'm not really sold that it appears like much of a sphere, and I've heard that if you want that you need to go with vector graphics (something I still know little on producing, but I know they are produced algorimically as opposed to bit-mapping). Is this true?

Another question here, and this may sound a little stupid - would there be any point in warping an earth-image around a sphere? Of course I know it's already a 3D object, but would I be able to make it stand out on the screen a little more around an imported vector sphere? I've been having a little fun with curving my personal logo around the earth :-), and if I could curve it around an even rounder earth (yeah, you know how Nigel's amplifier went up to 11?), that would be really cool!


Share on Facebook Share on Twitter Share on Orkut Share on Digg Share on MySpace Share on Delicious Share on Technorati
Top
 Post subject: Re: Do you need vector graphics to get better sphere illusions?
PostPosted: Wed Apr 02, 2014 9:53 am  (#2) 
Offline
Script Coder
User avatar

Joined: Oct 25, 2010
Posts: 4756
I think someone is confusing vector graphics and 3D, because I don't see what vector graphics would buy you in this case (vector graphics, as understood by Gimp and Inkscape, are purely 2D objects).

_________________
Image


Top
 Post subject: Re: Do you need vector graphics to get better sphere illusions?
PostPosted: Wed Apr 02, 2014 10:17 am  (#3) 
Offline
GimpChat Founder
User avatar

Joined: May 22, 2008
Posts: 5242
Location: Gimpville
You might want to take a look at the Map Object filter. There is a good tutorial here on Creating A 3D Bowl. A dome could be created the same way, just rotated upside down.

_________________
“If you reach for the stars, you just might land on a decently sized hill.” - Stuart Hill


Top
 Post subject: Re: Do you need vector graphics to get better sphere illusions?
PostPosted: Wed Apr 02, 2014 2:41 pm  (#4) 
Offline
GimpChat Member

Joined: Mar 24, 2014
Posts: 21
ofnuts wrote:
I think someone is confusing vector graphics and 3D, because I don't see what vector graphics would buy you in this case (vector graphics, as understood by Gimp and Inkscape, are purely 2D objects).

No, I understand what vector graphics are, if not all of their potential. They can be scaled from a small size to any size without quality loss on account of being algorithmically driven, as opposed to a fixed pixel map. What I'm not sure of is why that would matter with certain graphical effects, if it matters - I only know that (having used bump-mapping, the Warp tool, and the 3D Extrusion script myself) while googling up the desired effects, I landed on sites (probably somewhere on this one too before I joined) where the consensus was that some shaping effects were done better in vector. An specific example would be wrapping a layer around an object (and I went "WHAT, isn't that what the Warp tool is for?") Some have presented text fonts which aren't in Gimp and are in Inkscape, and even claim that there's effects available for working on text which isn't in Gimp. Anyway, I'd appreciate if you can weigh in on this question: are there any graphical effects (other than scaling your image real high and the availability of prettier fonts) which are naturally done better with Inkscape? It may save me a lot of time trying to do certain jobs with the wrong tool.


Top
 Post subject: Re: Do you need vector graphics to get better sphere illusions?
PostPosted: Wed Apr 02, 2014 3:05 pm  (#5) 
Offline
Script Coder
User avatar

Joined: Oct 25, 2010
Posts: 4756
Any geometric transform (not only scaling) looks better in vector: rotation, perspective because in bitmap you always end up interpolating pixels and this introduces blur even if its less visible than what you get when you upscale. But you have to be in vector.

But vector graphics aren't the silver bullet. The Bezier splines that are used under the hood have the very nice property to be preserved by affine transforms(*) but not all geometric transform are affine (for instance wrapping something on a sphere isn't).

(*) a cubic Bezier curve (the ones that are used in vector gfx) is described by 4 points. The image of the curve in a affine transform (ie, the curve obtained by applying the transform to the infinitely many points of the original curve) is the Bezier curve that is described by the four points obtained by applying the affine transform to the four control points of the original curve (which is much more efficient...)

_________________
Image


Top
 Post subject: Re: Do you need vector graphics to get better sphere illusions?
PostPosted: Wed Apr 02, 2014 10:05 pm  (#6) 
Offline
GimpChat Member

Joined: Mar 24, 2014
Posts: 21
ofnuts wrote:
Any geometric transform (not only scaling) looks better in vector: rotation, perspective because in bitmap you always end up interpolating pixels and this introduces blur even if its less visible than what you get when you upscale. But you have to be in vector.

But vector graphics aren't the silver bullet. The Bezier splines that are used under the hood have the very nice property to be preserved by affine transforms(*) but not all geometric transform are affine (for instance wrapping something on a sphere isn't).

(*) a cubic Bezier curve (the ones that are used in vector gfx) is described by 4 points. The image of the curve in a affine transform (ie, the curve obtained by applying the transform to the infinitely many points of the original curve) is the Bezier curve that is described by the four points obtained by applying the affine transform to the four control points of the original curve (which is much more efficient...)


So then, are you saying that the four points which control the curves are the limiting factor in vector image complexity, making it impossible for anything resembling nature (or too complexly surreal) to be produced in vector?


Top
 Post subject: Re: Do you need vector graphics to get better sphere illusions?
PostPosted: Wed Apr 02, 2014 10:56 pm  (#7) 
Offline
GimpChat Member

Joined: Mar 24, 2014
Posts: 21
GnuTux wrote:
You might want to take a look at the Map Object filter. There is a good tutorial here on Creating A 3D Bowl. A dome could be created the same way, just rotated upside down.

Here it is, with and without that Map Object:
Image
Hey, it really does go up to 11! :) Of course it distorts a little from the original image, but not nearly so bad as a Warped image in Grain Merge mode. Thanks for pointing that out!


Top
 Post subject: Re: Do you need vector graphics to get better sphere illusions?
PostPosted: Thu Apr 03, 2014 2:03 am  (#8) 
Offline
Script Coder
User avatar

Joined: Oct 25, 2010
Posts: 4756
MisterA wrote:
ofnuts wrote:
Any geometric transform (not only scaling) looks better in vector: rotation, perspective because in bitmap you always end up interpolating pixels and this introduces blur even if its less visible than what you get when you upscale. But you have to be in vector.

But vector graphics aren't the silver bullet. The Bezier splines that are used under the hood have the very nice property to be preserved by affine transforms(*) but not all geometric transform are affine (for instance wrapping something on a sphere isn't).

(*) a cubic Bezier curve (the ones that are used in vector gfx) is described by 4 points. The image of the curve in a affine transform (ie, the curve obtained by applying the transform to the infinitely many points of the original curve) is the Bezier curve that is described by the four points obtained by applying the affine transform to the four control points of the original curve (which is much more efficient...)


So then, are you saying that the four points which control the curves are the limiting factor in vector image complexity, making it impossible for anything resembling nature (or too complexly surreal) to be produced in vector?


The 4 points per se are not the limiting factor. Real curves are a sequence of individual curves. So in 2D can you can have very complex curves described by vectors (fonts are vector graphics...). But the fundamental principle of vectors is that you only deal with the 4 control points of each individual curve, the rest of the points is computed only when the curve needs to be rendered, and this assumption is only valid for simple transforms (which fortunately are the most frequent ones).

_________________
Image


Top
Post new topic Reply to topic  [ 8 posts ] 

All times are UTC - 5 hours [ DST ]


   Similar Topics   Replies 
No new posts Attachment(s) Gimp Vector Graphics?

17

No new posts using a graphics tablet

4

No new posts Attachment(s) GEGL vector layer plugin

1

No new posts What I think Gimp/FOSS graphics needs to go to the next level

14

No new posts Attachment(s) Glass sphere.py

7



* Login  



Powered by phpBB3 © phpBB Group