fastest way to code an one pixel sinus scroller

[Int42]

Hi,


what is the fastest way to code an one pixel sinus scroller on an Amiga500 OCS?


Should I create a standard scroller in a backbuffer with blit char, scroll the whole text and then get 1 Pixel columns of that and blit this in the shown screenbuffer?

[Antiriad_UK]

That's what I do

[Galahad/FLT]

Originally Posted by Antiriad_UK

That's what I do

Ditto

[Int42]

ok, thx then I'm on the road.

[Antiriad_UK]

The sine scroller from Sax Offender code is here:
https://github.com/jonathanbennett73...Offender/Miami

It's a vertical fill one but still does the 1 pixel column blitting. The copy to screen function is SIN_CopyScrollToSine

[defor]

Draw vertical lines with blitter, use 16 pixel pattern for every column. Pros: No need to draw characters to a buffer. As a bonus, it's easy to stretch text horizontally. Cons: 16 pixels in one draw call only.

[DanScott]

Originally Posted by defor

Draw vertical lines with blitter, use 16 pixel pattern for every column. Pros: No need to draw characters to a buffer. As a bonus, it's easy to stretch text horizontally. Cons: 16 pixels in one draw call only.

That's going to be horrendously slow though.

The quickest way to do a tall sine scroll (for example, font is 128 pixels high), is to plot the vertical edge points with the CPU, and then blitter vertical XOR fill. You can be filling the previous buffer while the CPU is plotting the next set of spans.

Another option is to use an animated font (sounds crazy, but does work.. have a look at Vision - "Can't Be" demo for a good example). This is probably the absolute fastest way possible.

[ Show youtube player ]

[Photon]

@Dan OCS/68000 can do a 270px high sine scroller in real time without filling. But I think Int42 is looking for a quick and easy way to write one, rather than how to write the fastest one.

ORing pixel columns from a scroll buffer is a bit simpler. You could fill the scroll buffer just to test your routine and see what height font it accommodates. 30px is possible this way.

[Estrayk]

OFFTOPIC to the coders.
I would like to know what has been the most difficult sinus-scroll to program that you have seen.
Med-Red Slayer/SCX or Corto/QTX ones maybe?

[defor]

Originally Posted by Estrayk

OFFTOPIC to the coders.
I would like to know what has been the most difficult sinus-scroll to program that you have seen.
Med-Red Slayer/SCX or Corto/QTX ones maybe?

Not sinus-scroll per se, but very impressive indeed: Mr. Mega Mind in various cracktros [ Show youtube player ].

[DanScott]

Originally Posted by defor

Not sinus-scroll per se, but very impressive indeed: Mr. Mega Mind in various cracktros [ Show youtube player ].

Have always been impressed with that one since the day it was released!

[Int42]

Wow awesome tipps here from you guys, thx a lot.

I'm interested in every 1 pixel Sinusscroll, easy to code or fastest.

How is this DoubleSinus coded? at minute 5:23
[ Show youtube player ]

[DanScott]

Originally Posted by Int42

How is this DoubleSinus coded? at minute 5:23
[ Show youtube player ]

Quite probably the same as a regular sine scroll, but rendering the strips into a wider screen buffer (enough to allow for plenty of horizontal scrolling per scan line)

So if you have a regular 320 pixel wide screen, and you want to apply 64 pixels of left/right sinus too, you'd render your standard sin scroll into a 384 pixel wide screen buffer

[Photon]

Any coder with some experience should be able to surpass Slayer's sine scrolls with today's knowledge available; as mentioned, there's some room to bandwidth ceiling using the simple approach.

The Paradox intro uses prepared code/data lumps to plot or not plot into a fixed pattern. This was first used by Fuzzac in Tropical Sunset/TSL and is not related to sine scrolls at all.

Originally Posted by Int42

I'm interested in every 1 pixel Sinusscroll, easy to code or fastest.

I don't think anyone has beaten Blu Sky yet. It's close to bandwidth ceiling.

[ Show youtube player ]

[Antiriad_UK]

Originally Posted by DanScott

That's going to be horrendously slow though.

The quickest way to do a tall sine scroll (for example, font is 128 pixels high), is to plot the vertical edge points with the CPU, and then blitter vertical XOR fill. You can be filling the previous buffer while the CPU is plotting the next set of spans.

Another option is to use an animated font (sounds crazy, but does work.. have a look at Vision - "Can't Be" demo for a good example). This is probably the absolute fastest way possible.

[ Show youtube player ]

Plotting with the CPU, must admit I didn't think of that. Are you like tracking the x,y of each set of vertical points and "scrolling" them left - I guess you can do that quickly? I did my big sine (with your help) just with the usual blitter columns and a big modulo to skip all the lines in between - never crossed my mind to try and plot with CPU

[Antiriad_UK]

Originally Posted by Photon

I don't think anyone has beaten Blu Sky yet. It's close to bandwidth ceiling.

[ Show youtube player ]

I think you said in the demo it was memory limited. Would love to know how it worked.

[ross]

Gha... I've a déjà vu

[DanScott]

Originally Posted by Antiriad_UK

Plotting with the CPU, must admit I didn't think of that. Are you like tracking the x,y of each set of vertical points and "scrolling" them left - I guess you can do that quickly? I did my big sine (with your help) just with the usual blitter columns and a big modulo to skip all the lines in between - never crossed my mind to try and plot with CPU

The perspective correct textured greets scroller in Colombia is based on the same technique

[defor]

Originally Posted by DanScott

The perspective correct textured greets scroller in Colombia is based on the same technique

I thought it used those "prepared lumps" Photon mentioned above Therefore similar to those crazy twisting scrollers from the past. Silly me.

[sandruzzo]

Remember to use Interleaved bitmap, that'll give you one blit pass for one line