gmane.comp.systems.sam-coupe

Re: Quick attempt at a scroller

2012-05-19T20:49:00

Quoting Balor Price <tobermory-9lc6n+nyeLEgpMukQH3Qe/XRex20P6io< at >public.gmane.org>:



Besides - if you needed to - it's always perhaps possible to load a  
few random ones from disk... :)

Re: Quick attempt at a scroller

Balor Price — 2012-05-19T13:47:02

Aha you flatter me I'm not /that/ old :)  (okay I'm 36)  Yes indeed it 
is Splitting Images - it's another puzzler without any huge technical 
challenges, so I know I'm working within my comfort zone.  I hadn't 
planned on making any changes with the personalities, but there are a 
couple of people who aren't immediately obvious, the speccy's version of 
Mick Jagger was terrible!

I'd agree with Si on this one - unless you totally optimise each sprite 
you aren't looking at many.  But there's masses of memory in the SAM so 
unless you're doing it to squash into memory-resident megademo, why not 
use that memory?  There's often a direct trade-off of memory-for-speed 
(eg unrolling loops, making several specialised copies of routines) so 
go for it.

Howard

On 18/05/2012 00:56, Tommo H wrote:

Re: Quick attempt at a scroller

Thomas Harte — 2012-05-18T16:20:07

That's really encouraging; I'm now obsessed with getting a complete
game working at 50Hz and in that context I think even just two or
three sprites would do it. The arbitrary 40% came from what was left
over in the current demo but it strikes me it's also very close to the
amount of time between line 192 and the next line 1 so setting a firm
limit in that range resolves all potential concerns about sprite
flicker (well, until I end up having to display a different three of
the four sprites I really want each frame, or whatever).

And once I have something working along those lines I can start seeing
how well the scrolling holds up with more realistic map data. I've a
few more optimisation ideas either way.

On 18 May 2012 09:46, Simon Owen <simon.owen-cmSZVAwZacRg9hUCZPvPmw< at >public.gmane.org> wrote:

Re: Quick attempt at a scroller

Simon Owen — 2012-05-18T08:46:39

Maybe around 4-5 sprites of 16x16 pixels?  IIRC my Pac-Man emulator has about half a frame to draw 6 sprites of 12x12, plus a couple of background tiles, including save/mask/draw/restore.  For an extra boost perhaps generate the sprite drawing code with knowledge of the data (like Chris did in SAM Defender) — bonus points if you prepare that at runtime.

It's nice to have some fresh on-topic content to read, so keep posting!

Si

Re: Quick attempt at a scroller

Tommo H — 2012-05-17T23:56:53

I read your blog a few days ago I think; it's Splitting Images isn't
it? Are you going to update the images or are (slightly) younger
people like myself going to have to find out what people like David
Owen looked like?

I'm working on sprites at the minute. Mostly mentally as my schedule
requires, but I'm just not sure what's realistic to attempt. If I
have, say, 40% of a frame to spend on it, how many sprites, and how
large, is it realistic to expect to be able to draw and un-draw?

Open middleware is definitely the thing to aim for. Besides anything
else, it'll make the assumptions and limitations plainly visible.

On 17 May 2012, at 22:36, Balor Price <tobermory-9lc6n+nyeLEgpMukQH3Qe/XRex20P6io< at >public.gmane.org> wrote:

Re: Quick attempt at a scroller

Balor Price — 2012-05-17T21:36:15

No apologies required, looks very good to me.  So if you're not in a 
position to turn it into a full game, would you consider turning it into 
a bit of middleware??  A bit cheeky I know but I'm sure it would find a 
good home somewhere, probably for many.  I made some experiments with 
'full screen' scrolling ages ago but tbh I didn't know what I was doing 
because the project was too large, here's a demo I found: 
http://scenedamage.com/cookingcircle/Random/AtomManBackup.dsk

I had a good time doing my last game and I'm halfway* through another 
one, having fun!  Developing with a cross-compiler/tools really makes a 
fantastic effect on productivity, and your tiles displayer/scroller 
sounds the business.  If you improve the algorithm, anyone using the 
routines would get a instant upgrade...

I believe I am rambling.
Howard


*OK, 20%

On 17/05/2012 00:23, Thomas Harte wrote:

Re: Quick attempt at a scroller

Thomas Harte — 2012-05-16T23:23:34

Apologies to all; I don't mean to treat this list as my own personal
development blog. However, here's a rock steady 50Hz, full-screen
rendition:

http://www.clocksignal.com/dropbox/scroller-fullscreen-50Hz.dsk

There are two versions on there and some quick relevant notes. Takeaways:

• the border masking is overdraw and I'd rather fix the problem by
just not drawing incorrectly in the first place;
• that being the case, I've only properly profiled the unmasked
version, and found it takes no more than 60% of the available CPU time
per frame in its current state.


On 15 May 2012 12:39,  <warren-DId23kj0q5b10XsdtD+oqA< at >public.gmane.org> wrote:

Re: Quick attempt at a scroller

2012-05-15T11:39:16

It's very cool to see scrolling that quick and smooth. :-) We just  
need someone to use it in a game now!!

Quoting David Sanders <dsuzukisanders-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org>:

Re: Quick attempt at a scroller

David Sanders — 2012-05-15T11:05:53


That is indeed some of the smoothest, fastest full-screen scrolling I've
seen on the Coupé to date though :-)

Re: Quick attempt at a scroller

Thomas Harte — 2012-05-15T10:32:38

Or, more likely, the sad realisation that I can't scale the thing to a
proper game is fast approaching...

For the record, this is it mostly at 25fps, running (essentially) full
screen with black guttering to hide the edge jittering of yesterday:

http://www.clocksignal.com/dropbox/scroller-big.dsk

It occasionally — and very obviously — leaps up to 50fps, having no
frame rate compensation mechanisms whatsoever.

I'm about 80% confident I can boost the speed to a reasonably constant
50fps, to drop to a constant 25fps with sprites and logic, for a
simple enough game.

On 15 May 2012 08:02, David Sanders <dsuzukisanders-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org> wrote:

Re: Fred 58: Sam Coupe book?

2012-05-15T08:50:47

oddly enough, having looked through FRED 58 recently looking at my old  
game, I came across that mention of the book and was curious as well.  
Like others, though, I have no idea what came of it.

It would be very cool to know what happened to it and see the  
articles/images collected. Maybe someone could finally piece it  
together to one of those 'on-demand' books?

Come to think of it, maybe that's something we should all start  
thinking about. It'd be a while before I had the time, but I wouldn't  
mind working on something like that since I've pieced together a  
couple of self-published books for my local writing group. It'd be  
pretty cool if perhaps a year from now, we could all have our own copy  
of a nice image-filled hardback about the SAM Coupe. :-)

Warren

Quoting Thomas Harte <tomh.retrospec-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org>:

Re: Quick attempt at a scroller

David Sanders — 2012-05-15T07:02:16

got real jobs to do? Excuses!


D

Re: Quick attempt at a scroller

Tommo H — 2012-05-15T00:42:29

No, there's no deltas in that sense. I just mean that I've got two
routines — DrawBlock and DrawBlank — each of which draws a whole tile
at (hl) and then returns. Betraying my dense level  of thought, they
both look essentially like this:

    ld (hl), b
    inc l
    ld (hl), c
    inc l
    ... etc etc ...

A quick mental calculation suggests I could go full screen again if I
switched to subverting the stack pointer. I wrote both by hand for
expediency, so it's no surprise they're suboptimal. I had a go with
drawing on the half-byte too for a proper pixel scroll but the
read/modify/writes killed performance (or, rather, my attempt at them
did). I guess either 16x16 tiles (for half as many shared borders) or
a vertical scroll would be potential solutions; a vertical scroll
would also fix my clipping without any trauma, I guess — I can easily
spare gutter space on either side.

I think it'd be nice to go full screen and properly clipped one way or
the other just to prove the point; I'm not sure I have

Re: Quick attempt at a scroller

Balor Price — 2012-05-15T00:09:55

Nice!

When you say the tiles are 'precompiled', have you used a sprite builder 
to print the deltas?  If so, I'm confused about the left-hand clipping 
if there's no real scrolling involved.  Surely you're not printing all 
those tiles each frame?

Howard


On 15/05/2012 00:32, Thomas Harte wrote:

Quick attempt at a scroller

Thomas Harte — 2012-05-14T23:32:05

It's exceedingly rough and a pretty simple effect that I'm sure has
been exploited a hundred times before but I thought I'd throw it up as
is as my part in maintaining the fantastic momentum we've had lately.

http://www.clocksignal.com/dropbox/scroller.dsk

It's explicitly not a mere demo effect; adding some sprites and making
a full game is a definite possibility, though you'd probably want to
drop to 25fps — there's loads of memory free even on a 256kb machine
and it's a full, genuine Mode 4 display. The map itself is just an
array of bytes in memory, so there's no real magic there and no huge
footprint. The workflow is based on a cross-platform map editor called
Tiled so that's easy to manage.

Beyond my painful lack of artistic ability there's no reason why it
need use only one tile. Ditto for proper clipping at the edges (which,
because the tiles are precompiled, would cost more RAM but not really
any more processing).

It was full screen until the last minute, when I discovered that some
sections of

RE: Fred 58: Sam Coupe book?

Simon Cooke — 2012-05-14T22:41:44

There should be a Fred file extractor I wrote a looking time ago somewhere on nvg, I think.

Re: Fred 58: Sam Coupe book?

david brant — 2012-05-14T20:51:31

If you set simcoupe's printer setting to print to file you then can use the print option in the mag part. then you will get a txt file.

other than that not heard anything myself.

david

On 14 May 2012, at 20:42, Thomas Harte wrote:

Fred 58: Sam Coupe book?

Thomas Harte — 2012-05-14T19:42:34

I've transcribed by hand (aside: is there an official transcription or
extraction of the various Fred articles anywhere?) Page 5 and about
two third of Page 6 of the editorial below. Did anything ever come of
the project?

"The big (BIG BIG BIG) news this month is that FRED will be publishing
a book all about the SAM Coupe sometime this year. it is hoped that
the book will be out in time for Christmas this year and will include
many stories from people in the SAM world about their version of the
SAM story, and any experiences they have had. Some stories are
expected to be quite technical, while others will prove to be very
funny. At the moment, Colin is getting in touch with everyone who
could have a good story (including me...) and someone who is not
involved will decide what will make the final version.

"I must stress that the book IS NOT going to be a cheap, photocopied
thing. It will be produced in the same way as any paperback novel, and
could find it's way into some computer shops around the country.

Re: border effects and tricks

Andrew Gillen — 2012-05-14T18:32:48

Simon,

This is excellent. Many thanks for putting the effort into this :) I fear it 
would have taken me sometime to get it right so I appreciate your deft work!

I shall have a play, and with a bit of luck the results of this will be 
present in my next project.

Cheers

Andrew

--------------------------------------------------
From: "Simon Owen" <simon.owen-cmSZVAwZacRg9hUCZPvPmw< at >public.gmane.org>
Sent: Sunday, May 13, 2012 6:31 PM
To: <sam-users-w01uWLE43QoXWF+eFR7m5Q< at >public.gmane.org>
Subject: Re: border effects and tricks

RE: Nyan Cat

Stefan Drissen — 2012-05-14T09:41:34

Are you saying that you've created a youtube client for the trinity
interface? ;-)

-----Original Message-----
From: owner-sam-users-w01uWLE43QoXWF+eFR7m5Q< at >public.gmane.org [mailto:owner-sam-users-w01uWLE43QoXWF+eFR7m5Q< at >public.gmane.org] On
Behalf Of Adrian Brown
Sent: zaterdag 12 mei 2012 02:40
To: sam-users-w01uWLE43QoXWF+eFR7m5Q< at >public.gmane.org
Subject: RE: Nyan Cat

Ok, so who is looking at Nyan cat, I reckon if I wasn't clever and said it
was 512kb only it would take about 2 hours, to be clever on a 256kb sam it
would take about 3-4. ;)

Re: border effects and tricks

Simon Owen — 2012-05-13T17:31:57

Try working from this: http://obo.homeip.net/BorderSAM.zip

The visible width is 36 blocks, and each out covers 2 blocks.  That
means 18 OUTs for the effect plus 6 more for invisible padding between
lines.

Adding a NOP after an OUT adds a single block for odd widths.  There
must be an even number of NOPs in each line for timing, with any spare
NOP added after the padding OUTs.  For every 2 NOPs added you must
remove an OUT.  Technically, 2 OUTs (8T total) should be needed per
extra block, but the ASIC contention alignment during the border write
rounds it up anyway.

I've kept the number of vertical lines the same as the original code,
though you can see more than that on a real SAM.  There's also some
interrupt re-entrancy with the current code, but it's allowed for in the
timing delays.

I've just had a chat with Chris on IRC and there may be SAM support in
the editor program at some point :)

Si