Rubber Vector
category: code [glöplog]
SCA:
That´s not "per scanline" ;)
In Raymarching "per scanline" would be rotating using fragCoord.y (UV.y)
You are rotating in object-space using ray-positions y-coordinate! While object-space is the same as world-space in your case, as you didn´t translate or manipulate your object in other ways before rotating it.
Sorry, didn´t want to let you look dumb or anything, but i guess this could confuse people if not clarified!
That´s not "per scanline" ;)
In Raymarching "per scanline" would be rotating using fragCoord.y (UV.y)
You are rotating in object-space using ray-positions y-coordinate! While object-space is the same as world-space in your case, as you didn´t translate or manipulate your object in other ways before rotating it.
Sorry, didn´t want to let you look dumb or anything, but i guess this could confuse people if not clarified!
and just as a bonus information about different spaces in graphics:
fragCoord (UV) is screen-space
fragCoord (UV) is screen-space
@Hardy.
No worries, thank you for the correction, and explanation. I have very little experience with shadertoy, and raymarching. So basicly i am dumb on the subject and anything helps me to gain a better understanding.
No worries, thank you for the correction, and explanation. I have very little experience with shadertoy, and raymarching. So basicly i am dumb on the subject and anything helps me to gain a better understanding.
@Hardy:
I adapted the code to now use uv.y
Would it be more correct "per scanline" now?
I adapted the code to now use uv.y
Would it be more correct "per scanline" now?
yes, if you want to let it look "oldskoolish" for sure!
but nowadays i would go with the object-space-rotation you had before...looks better in my eyes! also it´s a bit hard to rotate screen-space for one object only! ;)
but nowadays i would go with the object-space-rotation you had before...looks better in my eyes! also it´s a bit hard to rotate screen-space for one object only! ;)
Isn't that just because you rotate it too fast?
On the linked picture, there is a rotation of 1 degree at each line. If I make a rotation of 0.5 degree at each line, the bug disappears. So you're right.
But when I watch Octorubber by Blueberry, his vector has a wide twisting, so I guess more than 1 degree of rotation at each line.
Maybe I'm wrong or missing something?
But when I watch Octorubber by Blueberry, his vector has a wide twisting, so I guess more than 1 degree of rotation at each line.
Maybe I'm wrong or missing something?
So nobody has an idea?
Blueberry, Slummy, Paradroid, are you here??? :)
Blueberry, Slummy, Paradroid, are you here??? :)
It's hard to say from an unshaded still image, but my guess is that it is not a bug. It just so happens that the scanlines rendered with different time offsets create that shape.
In Octorubber, the time shift propagates at 200 scanlines per second, so 1 degree rotation per scanline would mean a full rotation in less than 2 seconds. I don't think it rotates that fast.
Even then, if you step through the youtube, you will see some frames here and there with artifacts very similar to the one in your picture.
In Octorubber, the time shift propagates at 200 scanlines per second, so 1 degree rotation per scanline would mean a full rotation in less than 2 seconds. I don't think it rotates that fast.
Even then, if you step through the youtube, you will see some frames here and there with artifacts very similar to the one in your picture.
Yepp, pretty much the same conclusion here :)
Could it be related to rounding errors? What happens if you rotate 0.5 degrees twice? Same exact result, or slightly different?
Thanks you for the answer! As I wrote on ADA forum, now it works fine. As you said, I rotated each frame too much.