xdaniel

Meh, decided to look for the CD-Rs just before going to bed, found them, burned the image to the disk, and... ugh... EDIT: Bugfixes, improvements, and a new debug menu!

It's like I'm subconsciously telling myself "screw checking on hardware, add moar stuffz!!"... because here's my home-grown font and text printing class in action: ...actually, part of the hardware testing involves, well, being able to read the FPS somehow after all. And the text printing functionality I've used before, ex. in the Parallax library, doesn't seem to play nice with the optimized DMA-based vertex loading, so I needed to write something myself that does support this correctly... Anyway, custom binary font format (+ converter written in C#, similar to the model format), print function takes printf-style format strings and parameters (%s, %f, etc.), text color and scaling can be set prior to printing. It's still a bit limited in features and clumsily put together, so this'll need more cleaning up and other improvements, and I want to add control codes for coloring, at least, at some point. Will look for those last few CD-Rs one of these days and (after, I dunno, three, four years?) buy a new spindle of them next month...

Ash/Satoshi is the idiot from the anime, who's based on the badass Red from the games, who in turn has had other incarnations in ex. the various manga. They're all distinct characters from one another, but all based on the original Red from Gen 1 and 2.

IIRC only the Debug ROM doesn't care if the DMA table is wrong, while every other version does, although I can't remember if that's been investigated thoroughly. And IIRC it's a pain to fix the table properly; don't remember the specifics, but spinout had posted about this before - something about the file order having to be correct, which would require rebuilding the whole ROM and fixing a lot of addresses?

Mine's in my profile, although I don't have my 3DS with me right now; lend it out to a friend for trading purposes (he's got one 3DS but Pokemon X and Y). It'll be another week or so, I guess.

Hrm, not sure if I want to mess around with the drive... My PAL DC already had problems reading GDs and CDs when I pulled the PSU board out to put it into my NTSC DC - the one I'm mainly using now - and that one's still working fine, so I don't want to risk messing it up... That aside, 58 to 62 FPS according to code and 54%/32 VPS according to emulator, at 64 models rendered, one texture load per group, all rotating independently: Hardware testing tomorrow, I guess... or today, considering the time of day. EDIT: Haven't tested on hardware yet, but have redone the model format. A simplified explanation of the format... There's two important parts to it now, header aside, down from three: texture data and geometry data. Texture data consists of a pointer table and the actual texture information (dimensions, pixel data, etc) the table points to. Geometry data consists of lists of vertices on a per-texture basis. So the first list contains all the vertices that use one texture, the next list contains all vertices using another texture, and so on. There's some superfluous data still in there that I'll optimize out for the next version of the format (i.e. when I actually try and port over EO1 dungeon maps), but for now it's working pretty well. That leaves the performance checks on real hardware as the next short-term goal...

I'm starting to hate inaccurate emulation... or more like, I'm starting to hate the fact that modern systems do not have accurate emulators, as an alternative to HLE and other shortcuts. The ca. 30 FPS I was getting apparently stemmed from me having nullDC's dynarec CPU core disabled, because I was (and probably still am) getting periodic emulator crashes when running my code using that core. The interpreter core works fine, but it's obviously pretty damn slow. Now, with the dynarec enabled, and when it decides not to crash, the code is reporting FPS fluctuating between exactly 58.82 and 62.50 - however, at the same time, the emulator's stats in the title bar report ~94-97% speed and VPS ("video frames per second"?) of 56 to 57! Ugh... Starting to think I'll need to waste another CD-R sometime soon, after cleaning what I have code-wise up some more, re-adding the performance meter I've had in the C-based project (modified slightly from the one in the KOS examples), etc. If a hardware test shows steady 60 FPS with 36+ models being rendered, and maybe independently rotated and stuff for good measure, then I'll just live with the slow interpreter. Still need to look into the vertex buffer stuff, tho. I will need to be able to render more and different models with varying vertex counts, otherwise this'll be somewhat pointless of a project... in fact, a floor in Etrian Odyssey 1 can be 35 by 30 tiles in size if I remember correctly - 1050 tiles, 1050 models wanting to be rendered! Of course, not all of them will need to be shown at once, having the code just render, say, 10 by 5 tiles ahead of the player (10 tiles straight ahead, with 2 more on each side) should be far more manageable. Side note: to my surprise, the dynarec-using instance of nullDC I left running while I was typing this post up didn't end up crashing. ...those crashes are fucking random and it's annoying! Because I DID get a crash before I started writing this! <.< Another side note: I'm still sick but I can't sleep, so it's 3:22 am and I'm sitting in front of my PC still with some fever. Fuck this cold or whatever the hell -.-

Quick update in the spoiler, going to bed soon, headache & slight fever.

So, despite having come down with a cold (or rather "tonsillitis" I think?), I couldn't help it but work on Frenda/DCEO some more... or, well, I dove into two big things: first, slowly rewriting the whole thing in C++ as opposed to plain C, to be more object-oriented; to have each 3D model as a separate object, for example, or to have multiple text printers, each with a different font, or somesuch. Should also help with game state management, I'd imagine. Also started spinning off the "Frenda" part of the project into a separate library, as that's basically gonna be the engine of the "DCEO" project. I also did shove the whole pre-C++-rewrite project onto my local SVN, so I can return to that at any time. The other big thing is fixing the performance issues with 3D rendering that I mentioned. Those have been fixed, but there's one tiny, tiny problem: every texture load pretty much negates any performance gain I got. Well, rather, the more texture loads I do, the slower rendering becomes again. Render the room 16 times, with one texture load per group per room, and we're at 30 FPS. Kinda like that... it's a bit difficult to describe and I'm still a n00b with the DC's PowerVR GPU. So, I guess I'll redo the model format to support one texture per model, which contains all the textures needed by it, which'll be "addressed" via the texture coordinates alone. ...or that's what I hope will work. I'm tired, not feeling well from being sick, so I'll think about that and work on it some more in a few days or so.

If that's what I think it is, it's because of a bug with any Pokemon in the Medium Slow experience group at level 1: http://bulbapedia.bulbagarden.net/wiki/Experience#Experience_underflow_glitch - Also the reason why there's no wild Pokemon at level 1 anywhere and why eggs hatch at level 5. Otherwise, good luck with the project! Front sprite looks alright, although I think the legs look a bit weirdly shaped. Not sure how to make them look nicer, tho, as I suck at custom sprites. Might just be me, too.

"#loghorizon 02, more world-building & explanation, a nicely tactical battle w/ fine BGM, and a mission for Shiroe & co. Off we go!" ...as I tweeted two or so days ago. I think I won't be disappointed by LH - it might not set the world of anime on fire, break new ground, etc., but I'm sure it'll turn out as a solid show. ...also, not yay for falling behind even further as I appear to have come down with a cold...

Quick update on a few things: MikadoXNA I haven't been working on recently at all, as Miasma and the awkwardly named "Project Frenda/DCEO" have taken priority over it. Miasma, in turn, I've been fixing things in here and there - ex. Shin Boukenjima should be working now - but haven't done much with otherwise. Currently, "Frenda/DCEO" is my main focus, and it's something quite ambitious: it's planned to become a Dreamcast "port", more or less, of at least the dungeon crawling and battles of the first Etrian Odyssey for DS, with elements of EO4 and whatever else I can think of that makes sense. Certain basic stuff needed prior to making the actual gameplay, etc., is done or in progress - ex. basic game state management, some helper functions for simplified controller reading and file I/O, and a custom binary 3D model format with parsers/loaders for textures, materials and geometry. There's a lot of optimization to be done, tho, as ex. rendering of the first room from SharpOcarina's demo scene 16 times at once drops FPS down to ~15. Need to look into DMA transfers, but there's a promising-looking article on the net about that, as well as some example code in the SDK. Speaking of the SDK, I'm not using an official, leaked one for this, but a free GCC-based environment with ex. the KallistiOS library. Well, next longer term goal is getting a dungeon floor from EO1 ported over and rendering on the DC - extract and convert the map tile models, get them loaded, arrange and render them according to the floor map, etc.

A few random notes: I think I'm kinda hyping myself for Log Horizon after that first episode. I really liked it overall, I really like the characters so far in particular (Akatsuki!! <3), or how the world actually reminds me of an (MMO)RPG thanks to such little things as monsters dropping loot, etc., etc. And again, dat OP! Sooo chock-full of Engrish but gets you sooo pumped! ...I just hope all this will not end in disappointment, but I'm positive that it'll at least not turn into a total trainwreck... one reason being that it's on NHK Educational TV, tho that's still no guarantee... Holy shit, Seitokai Yakuindomo is getting a second season! Hell yeah!! It took three years and a bunch of scattered OVAs, but it's happening!! Damn, I really am excited for this, and unlike Log Horizon above (as of right now anyway) I'm very positive that this won't disappoint me, as the first season was just great and the OVAs left me wanting more And in less positive news - well, for me personally at least - I am totally behind with Monogatari Series 2nd Season (2 episodes?), Prisma Illya (only watched 1 out of 10 so far), C3-bu (watched roughly half and it's finished airing by now) and haven't even started any of the shows of this season that I was interested in, besides LH...

Art and drawing is what I'm doing this week on the job orientation program class thingy I'm participating in, which has resulted in this so far:

Didn't miss it. Hell, it's underlined! And in RED! Didn't really think much of it, tho... even though I usually react somehow when I see it on my clock (thank god for the 24h system)