Thread: Beats of Fire
View Single Post
Old 07 January 2015, 17:08   #368
Registered User

leathered's Avatar
Join Date: Oct 2011
Location: UK
Age: 39
Posts: 304
At some point many moons ago I had to make a decision. Blitz imposes a limit on the amount of frames you can have in memory at one time, with a maximum of 1024 frames.

That may sound like a lot, but you basically have an object designation vs memory limit compromise to make. Generally speaking (and I'm not saying I go about things the best way) I've put this together using the maximum of resources available to me and made changes along the way to make it all fit in other words, I'm all for doing it the easy way (you may call that lazy if you will) but necessity is the mother of invention. In other words the games uses 2mb of memory because that is what I have available =).

For my 2mb machine I've not had to split the torso and legs into 2, when I looked at doing so for the player characters there was little benefit to be had (the legs are all different for the player punching frames, by the way I appreciate this is not so for all the goon types).

Even so, I'm utilising and shuffling every available frame designation constantly within the memory limitations. So we split the frames in 2 and suddenly there won't be enough frames for many different goons in memory at once. It may be a concession that's needed further down the line for whatever reason, but for the current development it isn't, and I stand by that.

Many wish to see an ECS version complete but as I've said I'm focusing on getting the code done on an AGA machine first, and adapting later. I'm not averse to receiving work intended for ECS but really, it is never going to work without the kind of consideration that's gone into the AGA version, which you're going to have a hard time realising if all you're trying to do is 'your own way'.

If you want to help really the best way to do that is by helping the current development for one thing you'll learn about what's really going on and how the development is being approached and be in a better position to tackle all this 16 colour business. This isn't a new project either, it's been ongoing for over 2 years that means for the most part that you need to try and adapt to it, rather than the other way around.

It means helping to convert something to a 5 bitplane graphic and then to 4 utilising everything we have available. As opposed to trying to do the whole thing in 16 colours without the copper.
Yes it looks okay but I can hear the Amiga massive complaining already...

Otherwise the only way this is going to work, is for the current stage sets to be adapted (meaning those that have already been converted). I can provide the gfx. The character palette will need to be considered first, then each stage in turn beginning with stage 1_1. The tiles have also been optimised for these stages. I'd rather be slotting everything in as such than doing all the mapping from scratch.

It seems the best thing to do then would be to cleverly tile the parallax background to be a part of the foreground and optimise. We can put copper colours behind that and it should look pretty good. But I'd like to have the tile sets without the parallax adjoining also just in case...

To be brutally honest, all this recolouring of everything with a converter is all very well and good, and some of the results are very promising but not the best way to fit the requirements of the engine per se, and therefore not entirely helpful. But please, enjoy fighting amongst yourselves if you will.

I could assign you to doing a stage each with the well established current palette, show how we are approaching the mapping and we can see how you do with 16 colours after that =)
leathered is offline  
Page generated in 0.08326 seconds with 9 queries