Developer's Blog

Dec 3, 2006 5:57 pm: When it's been four years since your last release, you tend to forget exactly all the things you did before. Luckily I save emails. I looked up the vendors we used for stickers and t-shirts for MindCandy Volume 1, and not only are they still around, but for the most part their prices appear to be the same. The best are the ones that never change.. because they don't have to. -- phoenix

Dec 3, 2006 5:57 pm: Transcoding, for those not familiar with video production, generically means transforming footage into some other format. In the broadcast world before the advent of HDTV, it meant converting PAL to NTSC or NTSC to PAL. There are two components to doing this: Alter the resolution (480 lines vs. 576 lines) and alter the framerate (59.94 fields per second vs. 50 fields per second).

Guess which one is harder?

Resolution is trivial; just use your favorite non-crap resize algorithm (I used Lanczos, although bilinear would probably have been fine because I was only resizing in one direction). But the framerate -- that's hell. In our case, we were going from PAL to NTSC, so that's 50 images per second to 59.94. Where do the extra frames come from? Do you just duplicate a few? Then horizontal scrollies stutter. Do you blend smoothly across time? Then motion starts to "shimmer" and blur. So what the hell do you do? Where do you get inbetween frames that look like what they would be if they had actually been sampled at that point in time?

Make them up, of course!

I came up with a process that uses avisynth and mvtools to analyze the motion between frames, and when a new inbetween frame is needed, it is synthesized from that motion analysis. 99 times out of 100 it works flawlessly and you'd swear the demo was created on an NTSC machine. For the one time out of 100 it does NOT work, you have to switch or blend, so I decided to blend in the middle of the frame right at the moment the next frame should switch in time. I then edited these two processes together, using the blend for sections where the motion had failed.

Yes, editing was tough, but it allowed me to splitscreen at times where necessary. For example Arte's end section has scrolling credits that look best with the motion method, while the not-quite-full-framerate 3d part looks best with the switch method. So you have both :-)

I DARE ANYONE TO MAKE A BETTER TRANSCODE. People who get the NTSC disc will be very pleased, and quite frankly, I think it's nearly untouchable (unless you want to spend six figures on a Folsom or Snell and Wilcox converter). -- Trixter

Dec 3, 2006 5:57 pm: The menus are finished! But, holy mother of crap, was it difficult. We took the money we had set aside for the menu budget and used it to purchase Cinema 4D with MoGraph, and we ended up doing the menus ourselves.

You know how hard menus are? It took Jeremy and I between six and eight weeks to come up with 1. a concept, 2. that fit the feeling and style of the demoscene, and 3. wasn't annoying or a hinderance to use.

In the end, we finally had all the projects done, and then we spent an entire weekend rendering everything. I set up a renderfarm of every machine over 900MHz I could find, including Jeremy's machines -- and his machines were in a different town so we had to set up networking/firewall/vpn settings between us :-) There were seven machines all cranking away two weekends ago, for a total of about 36 hours of rendering (ie. 252 individual computer-hours). It was cool, until I noticed a bug in 6 of the projects and had to feed them into the renderfarm again :-P

In other cool news, Jeremy found that his innovative "Jukebox" and "Random play without repeats" DVD authoring code DOES work with One Giant File(tm), so that means all the video will get an extra smidgen of quality.

Except State of the Art. That thing is hell, and looks like crap in MPEG-2 format even if you render it out straight from an emulator. The blue women at the beginning of the demo differ from their background in CHROMA ONLY, and MPEG-2 encoders assign much less importance to chroma than luma for obvious reasons, so that scene usually looks bad... and then of course the interference circle patterns jump in and all hell breaks loose with the motion estimation. Argh. -- Trixter

Dec 3, 2006 5:57 pm: You're hungry for an update, aren't you? I hear you. Well, from my end of things, I've finally completed the 5.1 surround mixing, so I can put my time toward helping elsewhere. Jim and Jeremy are making solid progress on the menus. Dan is helping out in gathering more commentary from the groups and volunteers. We expect around 20-30% uncommented that we'll probably fill in. -- phoenix

Dec 3, 2006 5:57 pm: Pretty much everyone who offered to help with menus either removed themselves from consideration for lack of time (or we couldn't meet their quote) so we've taken on the role of doing the menus ourselves. And by ourselves, I mean me and Jeremy. Pim did a fantastic job last DVD (I'm still amazed by his speed) but he's getting married in less than a month and he's never done 3D stuff before, so I decided to take it on. Because, you know, I'm somewhat under-utilized in this entire project :-)

So anyway, Jeremy saw a demo of Cinema 4D, a 3D program I had never heard of, and I checked it out along with some demo reels of its MoGraph motion plugin. I have to say, I was completely floored. There's still decades of study to do, but in only a few days I was making simple scenes and animating them. So I'll be working with Jeremy (who has artistic design license of the menus) to come up with something.

Because, you know, just because I can use the program doesn't mean I know what to draw. ;-) -- Trixter

Dec 3, 2006 5:57 pm: The second (and hopefully final) wave of 5.1 mixing is underway. I've learned a lot since I started the first wave over a year ago, so there's plenty of tweaking to be done. We're doing bitrate testing now so we actually have a complete, final set of audio and video.. for one demo, so far. :)

Oh, and the Lapsuus mix turned out fine.. pretty good actually. -- Phoenix

Dec 3, 2006 5:57 pm: Fun with MPEG-2 encoders: SOTA's "dancing on top of interference pattern" section is almost completely impossible to encode at the bitrate we have available on the DVD (8700kbps, if you're curious, due to the number of audio tracks). I found one encoder that can do it, but at the expense of the "all chroma intro hand dancing blue people" section. The chroma is all mangled. So I used my other favorite encoder, and that section survives but now we're back to the interference pattern problem. What to do?

Encode with BOTH encoders and take the best parts of each output, splitting on GOP boundaries. Which is a major pain in the ass, but representative of the level of dedication we go through for maximum quality :-)

Along similar lines, I found yet another encoder that analyzes each frame and if it deems the content to be progressive, marks it as progressive; if interlaced, marks it as interlaced. This would give the best possible results from home theater deinterlacers, but there's one snag: It almost completely breaks most software DVD players. Since we know that many people buying the DVD would like to watch it on a computer, we can't use this mode. But don't worry, the quality is already extremely high, and to be honest, the progressive/interlaced flag thing would have been more for bragging rights than actual performance. :-)

The things we go through for quality. -- Trixter

Dec 3, 2006 5:57 pm: Grr.. it's been hell trying to take a high-quality WAV (Lapsuus, in this case), and make it play two seconds faster to sync with a video capture.. without drunken hiccups popping up all over the place. If you thought the rhythm in the 5.1 mix of Love on the preview disc sounded a little quirky.. you picked up on the same issue. Basically, I can choose between muffling and grinding, or stuttering and skipping. Sigh.. maybe I can get Trixter to stretch the video out two seconds longer. ;)

And.. I turn 30 tomorrow. I was 27 when we started working on this volume. Oh well, when I retire, I'll have more free time!
-- Phoenix

Dec 3, 2006 5:57 pm: With submissions from just two groups, 20% of the audio commentary is in. Cool. Obviously we want more. :)

Meanwhile, 5.1 channel mixing is almost done, I just have to get full-quality tracks for the newer, non-MOD demos.

Rest assured, there's a darned good chance you'll see Volume 2 released this year. :)

-- Phoenix

Dec 3, 2006 5:57 pm: Raw Sewage! Project work was halted this past week (Happy Holidays!) while my basement filled with raw sewage and I had to get the problem taken care of. Fun? Not really. Video exists if you want to get grossed out. -- Trixter

Dec 3, 2006 5:57 pm: So long a project and so little updating...! I'm not dead but I've come close; it's a simple sad fact that family takes precedence over hobbies. That being said, family life is calming down and I've returned to working on the project each night.

My latest piece of fun? Trying to figure out how to encode State Of The Art. Try this little test: Capture 50fps SOTA from your favorite A500 or emulator, then try to encode it via any MPEG-2 encoder and see if you can get the "circle interference" sections to encode without artifacts. Guess what? You can't! Two moving concentric circle interference patterns are pretty much a test pattern for "what MPEG-2 was *not* designed to encode".

The only way to get certain parts of SOTA to encode properly is to encode the demo as Half-D1 (352x576) so that you can get more out of the same bitrate... but of course we can't really do that since single-pixel detail may be lost. Argh! At least the 640x256 title screen looks rock solid. (Yes, that image is an analog capture encoded to MPEG-2, NOT an emulator screendump!) -- Trixter

Dec 3, 2006 5:57 pm: Preliminary captures are done!!! Yes, the project has taken 2 years to get to this point, but they're all done and things are starting to roll down the hill. The only demo left to include is a Madwizards PPC demo and I'm already in contact with them to get something. -- Trixter

Dec 3, 2006 5:57 pm: the screen ad for asm.. the dvds arrived at the party on friday night. phew!

-- phoenix

Dec 3, 2006 5:57 pm: The preview disc is done! I uploaded a 1.2GB ISO (twice -- had to squash a bug) today across the country in about 4 hours. Funny to remember the 80s, when downloading half a Meg would have taken as long.

With any luck the DVDs should arrive at Assembly on opening day, but mail carriers being who they are, one never knows. In any case, Abyss has kindly offered to distribute the ISO on the party network, so if you have a burner (or you don't mind using a software player) you're in luck.

By the way, the physical preview disc is PAL, but we will have an NTSC ISO available for download. Of course, these being Amiga demos, PAL is preferable.

Mindcandy 2 will be a "tech demo" in its own right, with some features pushing the DVD spec beyond its perceived limits. We hope you enjoy the taste of things to come. We'd love to hear your feedback. -- Jeremy

Dec 3, 2006 5:57 pm: hmm.. it seems we haven't updated this blog much lately, because either: 1) we haven't been doing anything, or 2) we've been too busy to update the blog. not much middle ground lately.

anyway, as trixter wrote, we're fighting to get a preview disc done. you mainly have jeremy to thank for that. he's assumed most of my whip-cracking duty. :) he finished up the menu a while back, a "cute" concept just for the preview disc. my 5.1 mixes are done, just a tweak or two tonight and final encoding. meanwhile, two weeks to the party and we still lack any kind of packaging for the disc! we're quickly debating whether to rush a small job thru our previous disc maker, or just hack some copies up ourselves. then of course, rush delivering the discs to finland. all to get shiny circles in some people's hands while the rest can download it a few days later. call it a motivational exercise!

btw, none of us can make to asm, unfortunately.. a seminar for mindcandy/video capturing would've been cool, but we just didn't prepare for it. check out the freax book seminar instead. :) -- phoenix

Dec 3, 2006 5:57 pm: I love archived USENET messages, and I love people who still reply to them even more. While we've already bought the rev b. board and it is on its way, I found the following accurate assessment of why the accelerator won't work in a rev D motherboard:

(begin quote)

My MkII was 040/40, and required "internal clocking". The internal clock signal is taken from U104 (50 MHz crystal). It was not installed in my A4000cr revD (it used the "external clock" on the A3640), so when I replaced the A3640 with the CS MkII, the computer refused to work. I just got a blank grey screen.

(end quote)

I asked this person if he ever found out how to fix it. His reply:

Soldering the crystal into U104 solved the problem.

Sorry, but I'm not that comfortable soldering a 50MHz crystal onto the motherboard (and a rare one at that). :-) So I'm glad we found a new motherboard.

During this downtime, I've been flexing my democoding muscles for some hardware that I don't think was ever given a proper demoscene treatment: 8088 with CGA. I have no idea why coding a demo for 8088/CGA is fun, but dammit, it is... Anyway, it's just some hobbyist dorking around -- when the new motherboard comes, it's back to capturing. -- Trixter

Dec 3, 2006 5:57 pm: If anyone's curious, I'm up to Human Target in the 5.1 mixes. My techniques are improving, I'm particularly happy with certain parts of Voyage and Hardwired.
-- Phoenix

Dec 3, 2006 5:57 pm: New rev b motherboard has been located and purchased. Yay! When I get it in the A4000 I'll update the blog. -- Trixter

Dec 3, 2006 5:57 pm: Put the A4000cr Rev D. machine all back together -- had a few "yellow screens" which gave me a scare but it turned out to be the ram SIMMs not fully in their sockets. (In fact, the clips holding them in are the cheapest plastic I've ever seen and I wonder how they were ever expected to work!). But the a4000cr is working.

The A4000 rev B machine is waiting for a new motherboard, and I think I've located one. Stay tuned. -- Trixter

Dec 3, 2006 5:57 pm: Very very bad news... it seems that my new "stock" A4000 that I was going to transfer the accelerator into is the last rev motherboard ever created, the A4000cr... which is sufficiently different enough that it doesn't support accelerators with external clocks. Guess which kind of acclerator I have :'-(

Crud. This halts the project until I can find an A4000 rev "B" motherboard. -- Trixter

Dec 3, 2006 5:57 pm: I've never taken an A4000 apart before. Let the "Trixter is an idiot" jokes begin, because I couldn't figure out how to get the faceplate off without breaking one of the tabs, so I gave up and will try again tonight. (Removal of the faceplate, as far as I can tell, is necessary to remove the drive bracket, which is necessary to remove to get to the CPU board.)

When I get it completed, though, I am pretty sure the older A4000 will be resold on ebay to raise some money for the project... this is because, in addition to having slightly dimmer video and broken audio on the right channel, it is a Rev B motherboard which isn't nearly as nice as the Rev D motherboard I have in the recipient machine.

Well... maybe I should hold onto it as a spare ;-) -- Trixter

Dec 3, 2006 5:57 pm: Problem with Concrete is solved.

Now I have to rip my A4000s apart and combine the best of each to make a single decent machine. I have Tint to thank for this. Here's the skinny: A good friend of mine, Ed Bravo, graciously donated his entire Amiga stash (including some rare Amiga hardware that I still don't know what to do with, like a light gun and various kick roms) to the project and in it was our fabled '060/60 accelerator installed an an A4000 with 80MB of RAM (64 on the accelerator of course). However, I noticed that the right audio channel was significantly distorted. My brother and I replaced the surface-mount capacitors but it still sounded awful, so we bought another A4000, stock, with 16MB fastram, to get decent audio out of demos. (BTW, I don't hold a grudge against Ed -- he is deaf in one ear, he could never have known! :-) So, this dual-A4000 arrangement has worked well for a while; I capture on both and edit the results together.

Until now. Tint runs so slowly on the stock A4000 with good audio that some notes are actually delayed during playback! So I am going to take the CDROM, SCSI hard drives, and '060 accelerator out of the broken A4000 and transplant it into the one with good audio tonight.

Wish me luck! -- Trixter

Dec 3, 2006 5:57 pm: Having major problems with Concrete. If anyone can help, please read the discussion over at pouet: http://www.pouet.net/topic.php?which=1972 -- Trixter

Dec 3, 2006 5:57 pm: Sorry for the lack of updates; for three days (I only have an hour or so a day recently to work on this) I have been struggling with Captured Dreams -- it runs slowly on a stock A4000/40, and runs great but completely out of sync on a '060/60. On top of that, on BOTH machines the demo has many bugs which I am conveniently editing out so that the demo looks better than it really is :-) I don't consider this harmful -- I am trying to merge my captures together in a manner that represents how it is SUPPOSED to run.

Hope this doesn't raise anyone's hackles. At least the video and sound are 100% Amiga and not emulated. -- Trixter

Dec 3, 2006 5:57 pm: Movement has been steady for the past three weeks. Yay!

I caved and rendered out a section of Nexus 7. I feel so... dirty. I did this to get past a scan converter bug that simply wouldn't capture a certain part properly. Don't worry, I did extensive split-screen tests and adjustments to match it up to the captured footage down to the pixel. You won't notice it (whereas you WOULD have noticed the scan converter bug). -- Trixter

Dec 3, 2006 5:57 pm: As of right now, all OCS/ECS segments (minus easter eggs) are fully edited, yay! A lot of time was spent removing noise from the A500's relatively low RGB output. All three of my A500s have this 'problem' so I'm beginning to think it's not a problem at all, but a design issue with the A500. In any case, they're done, and I only had to resort to using emulation once for a 20-second sequence -- the rest 99.999 percent is pure Amiga. -- Trixter

Dec 3, 2006 5:57 pm: Captures are coming along nicely on the A4000, if by 'nicely' you mean that Love won't run all the way through on anything I own, and that most AGA demos made past 1997 look like crap without an '060. :-) -- Trixter

Dec 3, 2006 5:57 pm: So I dug around and found 4x4MB SIMMs that fit (loosely!) in the A4000. So I now have a stock A4000/40 with 16mb fast and 2mb chip... and damn if some demos are decidedly less than impressive. It looks like I'm going to be capturing everything twice: Once on the stock A4000 to get clean audio, and again on the '060 acelerator for the best framerate (you'll recall that the '060 machine has bad audio), and then use both captures to create the final edit.

Why not just move the '060 to the good machine? Because there are a few demos that will NOT run on the '060 and I want to have both available. I also don't want to damage the '060 moving it all over the place, since they are rarer than hen's teeth at this point in the game. -- Trixter

Dec 3, 2006 5:57 pm: I am alive! With a particularly bad work month out of the way, I am back at work on the project. The second A4000 has been unboxed and it runs perfectly and, best of all, the sound is 100% pristine and has no flaws. Woohoo! Unfortunately, it also has no fastram, which means it runs Friday at Eight and Nexus7 and... not much else. Anyone know what kind of SIMMs a stock A4000 takes? -- Trixter

Dec 3, 2006 5:57 pm: Not much work done this weekend because of my real-life job obligations. I did manage to come up with a reliable and swift backup/sync system, so we are working with a safety net now :-O I also spent about 30 minutes trying to clean up a noisy capture (and the noise wasn't my fault; there are simply some bad colors in some demos that produce noise with every scan converter I've tried). -- Trixter

Dec 3, 2006 5:57 pm: I'm no longer sick! I could say "Production resumes!" except that a lot has already been going on. Jeremy has created a DVD structure with features that has never been seen in the DVD world before (like a working flexible jukebox!!) and Andy has been hard at work making 5.1 mixes of all the audio. And I received a package today that is probably our backup drive, which means I can capture and edit with a safety net!

Ah, what the heck: Production resumes! -- Trixter

Dec 3, 2006 5:57 pm: I have pneumonia. -- Trixter

Dec 3, 2006 5:57 pm: Just finished the 5.1 mix of Condom Corruption, and it is sounding go-o-od. It's taken me a while but I'm learning techniques as I go along. Now I just gotta mix the intro and sync the parts and I'll have my first demo track done for the crew! :) -- Phoenix

Dec 3, 2006 5:57 pm: I have the flu. 101 degree temperature. Don't expect any updates for a few days. -- Trixter

Dec 3, 2006 5:57 pm: No updates for a while because I'm sick. But all but one of the OCS demos have completed final edits and are awaiting approval. Looks like next week I'll be cracking open the A4000 box... -- Trixter

Dec 3, 2006 5:57 pm: Final edit of Mental Hangover is done. I broke down and used emulator output for the checkerboard (top part is still real, rest of the demo is real). Sorry about this but I need to move on to other stuff... I'll come back to it if I have time. -- Trixter

Dec 3, 2006 5:57 pm: Final edit for State of the Art is finished! Preview version goes off to Andy so he can do a 5.1 mix.

Mental Hangover is going to be the only demo that uses emulation, since I just can't get the endpart to work. So 99% of the demo is real, and the endpart will have the bottom half emulated (the top part and music are real). -- Trixter

Dec 3, 2006 5:57 pm: All floppy drive problems have been fixed, and some new dumps of demos have fixed other problems. So where are we on the ECS list?

Mental Hangover / Scoopex - Works until the endpart, where the bouncing balls section is totally messed up UNDER the logo, but the music plays fine and it reboots when done.

...and that's it! If I can't figure that one out then I'm going to have to emulate just that one part and merge it into the edit, sorry :-(
-- Trixter

Dec 3, 2006 5:57 pm: Thanks to Phoenix (new disk image) and a suggestion from Tom Roger Skauen (1mb chipram) did the trick, so we have Hardwired working. -- Trixter

Dec 3, 2006 5:57 pm: Just got Voyage working. I am definitely having drive issues... dammit.

I have to say this: I must have missed Voyage (was still only PC scene in 1991) and I have to say: Holy mother of crap, that demo was ahead of its time. There are some disposable parts (like the robot on two legs, that was obviously prerendered) but the texture mapping and ordered dither polys? God, I love that stuff.

"This voyage will never stop." Prolific and hokey at the same time :-) -- Trixter

Dec 3, 2006 5:57 pm: So, working out some disk drive troubles, we are left with:

Megademo / Red Sector Inc. - Works, although I had to pause and unpause a scroller to get one part loading, and endpart questionaire crashes when you pass/fail quiz

Mental Hangover / Scoopex - Works until the endpart, where the bouncing balls section is corrupt and it crashes soon afterwards

Hardwired / Silents & Crionics - Raytraced toilet is totally screwed up; otherwise fine.

Arte / Sanity - 99 percent perfect (one character of the endscroller font is mangled)

Jesus on Es - Crashes about one minute in, cannot find a cause

...and I haven't tried Voyage yet. -- Trixter

Dec 3, 2006 5:57 pm: OH MY GOD IT IS A DISK DRIVE ISSUE!! World of Commodore didn't work at all. Not a single boot. Formatting the disk (on A4000) produced no errors, yet didn't boot. So I format the disk on the A500, transfer to the A4000, write the DMS file, still didn't boot. So, taking the advice it is probably a track alignment issue, I did the following: Hook up 2nd drive to A500; copy sanity-woc92.dms and DMS itself to floppy on A4000 and stick in second drive of A500. Issue the following at CLI: "df1:dms write df1:sanity-woc92.dms to df0:" to write the dms file ON the drive it will be booting from.

And now I am watching WOC92, and loving the (first part) music. Goddamn disk drive!!! I never got any errors at any point, and yet it must be a subtle alignment error. Oh well, at least I have a 2nd disk drive so that this procedure is even possible.

In other fun discoveries, the original unpatched State Of The Art does not run on my A500, while the patched one ("SKID ROW DEBUGGED THIS") does. Joy. -- Trixter

Dec 3, 2006 5:57 pm: No work on the project last night or tonight because I'm helping out fellow history preservist Jason Scott with his BBS documentary. I highly recommend you check out www.bbsdocumentary.com. Right now. -- Trixter

Dec 3, 2006 5:57 pm: Spent tonight looking into noise reduction. I love ViXen, but I really don't want to set up Premire 6.5 again, as Premiere Pro makes it look downright clunky. Why oh why isn't there a compatibility mode for older plugins!?!

Anyway, I should have some pictures up soon. Remember that the MPEG encoder sees stuff you don't, so even on "perfect" captures I need to do some noise reduction. -- Trixter

Dec 3, 2006 5:57 pm: Many thanks to everyone who has given me suggestions; while I still have issues, I decided to increase the good mojo by rendering out a demo that has no problems running at all, State of the Art. A teeny tiny bit of color correction, a moving intro caption, massage the audio a fraction, et voila: Our first demo is done! To celebrate, I'm opening one of my Sanbitters -- bitter soda that resembles seltzer, cherry juice, and grapefruit juice. Kick ass.

Many other demos will require some light noise reduction due to a particular combination of colors they use (I don't know what it is about scan converters but there are some colors they simply can't handle), and all of them need 5.1 remastered music, and I'm sure there are a ton of more problems to come... but I'm very happy with this first milestone. -- Trixter

Dec 3, 2006 5:57 pm: Okay, here is a rundown of the troubles I have been having with the A500. My A500 is a real PAL amiga, has 1MB fat agnus, configured as 512chip/512fast (although set to 1MB chip did not change anything). Here is a quick rundown:

Megademo / Red Sector Inc. - Works, although I had to pause and unpause a scroller to get one part loading, and endpart questionaire crashes when you pass/fail quiz

Mental Hangover / Scoopex - Works until the endpart, where the bouncing balls section is corrupt and it crashes soon afterwards

Enigma / Phenomena - Seems to work perfectly (some endpart oddness but cannot tell if that is part of the demo)

Hardwired / Silents & Crionics - Starts with Fake Fast Located (LOL!!!), but music after title sequence is corrupt (just clicking). Demo runs; just no music. Raytraced toilet is totally screwed up. Various parts bug throughout.

Human Target / Melon Dezign - 100 percent perfect

World of Commodore / Sanity - Does not boot :-(

State of the Art / Spaceballs - 100 percent perfect

Desert Dream / Kefrens - 100 percent perfect

Groovy / Lemon - 100 percent perfect

242 / Virtual Dreams - 100 percent perfect

9 Fingers / Spaceballs - 100 percent perfect

Arte / Sanity - 99 percent perfect (one character of the endscroller font is mangled)

Guardian Dragon - Crashes after skullbat picture -- UPDATE: Loaded!!! holy crap.

Jesus on Es - Crashes about one minute in, cannot find a cause

Mobile - Destination Unknown -- runs okay for a minute, then crashes

So, does all this sound like 1. screwy disk drive, 2. a broken mainboard, 3. disks too old, 4. what?? -- Trixter

Dec 3, 2006 5:57 pm: Guardian Dragon is working!! I guess if you reboot enough times, set the machine to 512chip/512"fast", and blow out the innards with a can of compressed air, things work :-) But others are severly bugged, so I'm going to post an update soon with some test results in hopes that someone can clue me in what is going on... -- Trixter

Dec 3, 2006 5:57 pm: Oi! Join the discussion! -- Trixter

Dec 3, 2006 5:57 pm: So all three Amigas I have hooked up here have the "too dark" black problem. A better term for it is an incorrect gamma curve, probably done on purpose to support the monitors that were being corrected... because I hooked up a Commodore 1084 monitor to the A500 and they sure look a lot lighter (after fiddling with monitor controls).

So, I'm going to fix the gamma problem in post. Luckily, Premiere Pro has new filters that can fix this that weren't present in previous versions. And I'm going to fix levels as well, because I have found through testing that when you turn the gain up too much on the scan converter, the colors go hot on you! Meaning, if too much gain is applied, the grayscales tend to go from a nice neutral gray (leaning ever so slightly toward blue) to red-tinted grays, which sucks. It's almost like the overal temperature is going from 9300 kelvin to 6500... and if I went to television engineering school I could probably confirm/deny this but for now I'm letting it go.

So I'm going to fix stuff in post, and I'm already miffed about the banding present in 8-bit-per-channel raw files. (Mind you, 8-bit raw is still WORLDS better than the MPEG-2 I-frame stuff we did last time.) So I should probably capture in 10-bit, do my post-processing, and then render out to 8-bit for encoding (because I don't know of any software encoder that will accept 10-bit input, even though MPEG-2 encoders have a 10-bit resolution on the DCT... argh... don't get me started). Which means, 10-bit capture is going to eat up 25MB/s, and I am definitely going to have to cobble together some sort of backup solution for this. Blargh! But I still love it! -- Trixter

Dec 3, 2006 5:57 pm: Family problems, no work done. Trying to decompress.

Music: Thumpa by DJ Micro -- Trixter

Dec 3, 2006 5:57 pm: Had family stuff to do last night, so no progress. But I'll say this: The project is on my mind a LOT -- as evidenced by me entering this blog entry while I'm still at work. This is a good thing! -- Trixter

Dec 3, 2006 5:57 pm: Hooked up the donated NTSC A500 from my friend Ed Bravo, and I am relieved (?) to see that it, too, has a black crush problem (the bottom 8% of luminance values all crush into black). I guess this was a problem with all A500s, yes? Also, tested the audio and was happy to see that it has good audio, just like the other A500 (and unlike our first A4000).

So, if both A500s have a black problem, do I capture demos on an A500 to be authentic, or do I capture them on an Amiga that fixes the problem? Please email me with advice...

Can't find a goddamn OS boot disk so I can't run degrader. I'll hook up the A4000 to build one -- tomorrow. Time for bed :-( -- Trixter

Dec 3, 2006 5:57 pm: I've been sick for the last two days... hopefully will progress more tonight. I've found a few demos that run on the damaged A500 (Human Target, Groovy, SOTA, Desert Dream, some others) so I'm going to capture them and worry about noise reduction later (there is, unfortunately, a bit of noise in some of the static solid color screens).

I also have an NTSC A500 that I want to hook up and test with Degrader to see if 1. the timing is the same (probably not), and 2. if it has the crushed blacks problem the current A500 has. -- Trixter

Dec 3, 2006 5:57 pm: So I encoded video for three hours, had Jeremy download it at his place for another three hours, and... the output was all mangled thanks to Cinema Craft Basic not being able to handle Premiere Pro + Black Magic Design codec. Argh! -- Trixter

Dec 3, 2006 5:57 pm: So how do you cope with a scan converter that can't quite exactly 100% perfectly match up with the Amiga's RGB output? When it slightly mangles one frame every ten seconds? Simple: You capture the demo TWICE, and wherever there is a mangled frame, take it from the second capture where there is a 1-in-250 chance it also won't be mangled.

So today's dilemma is that the A500 I am working with drops off the last notch of luminance. So where you would normally see values 0-15, 2 is very dark, and 1 is the same as 0 (black). The A4000 doesn't have this problem -- but then again, it may not run the demo 100% exactly as intended. So what do I do? Email me. -- Trixter

Dec 3, 2006 5:57 pm: The line-level converter came today and it did the trick! Without it, levels were too low and during recording I was saturating about 11 or 12 bits of all 16 bits of dynamic range. But with it, I am saturating 15-16. Of course, this raises the noise floor to around -60dB, but since 8-bit audio cannot mathematically exceed a dynamic resolution of -48dB, I'll just do a noise cut at -48. Woohoo!

Yes, I know I could have gotten a mixer, but I already have a mixer for the audio monitors. Which seemed like the right choice until Dan reminded me that a real mixer would have helped us the next time we record commentary for a future volume. DOH!! Oh well, we'll buy one when we need it.

In other news, the A500 Ville sent us is definitely bad. It has some sort of memory problem :-( I'm a little miffed because I JUST got the damn setup working, but it's not his fault and I'll see what I can do about getting another one.

Last night I couldn't sleep due to a levels issue. MPEG-2 (601) has a "visible" luminance range from 16-235. When I put a 0-255 grayscale ramp in the timeline, it became banded due to the conversion of 0-255 -> 16-235. You know something? This blows. I can capture and work in 10-bit to avoid banding (which is what occurs when there is roundoff during the conversion) but guess what? I can't pass 10-bit to the MPEG-2 encoder, only 8-bit. So any demo I try to capture with 256 distinct levels of gray is going to end up with 220 instead, whether it likes it or not. (This is made especially ironic by the fact that most decent MPEG-2 encoders allow a DC coefficient resolution of 8, 9, or 10 bits! If only I could feed the encoder 10-bit, I could take advantage of 10-bit DC coefs...)

Music: Echo's soundtrack to Jesus on E's -- for the first 40 seconds, because it crashes after that. And why does that demo take so damn long to start running? -- Trixter

Dec 3, 2006 5:57 pm: Okay, I have been working next to my A500 for over an hour and the constant clicking of the (idle!) floppy drive is driving me nuts! How did you oldskoolers ever stand it?

Then again, I guess you don't hear the floppy drive at a loud demoparty :) -- Trixter

Dec 3, 2006 5:57 pm: Working on the project. "DO YOU LIKE FILLED VECTORS?" my video monitor is displaying right now.

HELL YES, my friends. I DO LIKE FILLED VECTORS.

Music: Future Sound of London -- Trixter

Dec 3, 2006 5:57 pm: Ordered a line-level converter and had them ship it 2-day air. I want to have some captures done before the weekend so that I can start reading/answering my email again :-)

BTW, one of the forces moving my lazy ass again is the knowledge that I am okay with the end result not being 100% perfect -- that 99% or even 98% is okay. Not many people know this outside of the project, but when I was working on the first Mindcandy, I was an insufferable perfectionist -- I would tweak a particular sequence for days and days. Some of the results were worth it (all of the tweakmodes required extensive post-processing and a little ingenuity to not look like ass) but it took a toll on my family and my day job. I was getting 4 hours or less of sleep a night for several nights in a row, ignoring my wife and kids half the week... it wasn't good. I realize now that only about 1% of the people who will view this next DVD will be able to see that kind of effort, so I'm going to just accept that 1% of you all will be pissed and the other 99% will love the DVD all the same.

I write this because I have just demonstrated the Blackmagic Design codec's colorspace conversion on my production monitor with a 0-255 grayscale ramp, and sure enough it collapses to 16-235 -- which isn't a bad thing except it collapses slightly ungracefully and there is banding. Now, which demos use a 0-255 grayscale ramp with all 255 grays displayed at the same time? None of them, but it still bothers me. Part of me wants to switch to 10-bit mastering to get around it (currently at 8-bit mastering), but that will put even more strain on the system, take up even more space (datarate is already 20MB/s), and make the resulting .AVI files incompatible with applications outside of Premiere. So I'm weighing the good against the bad, and because of the massive slippage and budget constraints, I'll stay in 8-bit.

Besides, it helps immensely that I am capturing directly to 16-235 thanks to proper calibration of the scan converter. Thank you, calibration unity!

Music: None at the moment. I feel strangely neutral. -- Trixter

Dec 3, 2006 5:57 pm: Well, this evening I taught myself what industry veterans went to school to learn:

- XLR ("pro") is used for loud things (+4dB)
- RCA ("consumer") is used for quiet things (-10dB)

This is why audio is too low. I'll order some line-level converters tomorrow -- wait, today. I can't believe it's so late -- hope I can get up for work tomorrow.

I also drew a few test patterns in The Gimp and ran them (and an "official" test pattern supplied by Black Magic) through Premiere Pro's lovely new monitors and finally FINALLY learned how the Blackmagic design codec (BMDC) is translating colorspaces AND how to read a PAL waveform monitor! For anyone wondering, all signs point to the following:

- PAL waveform monitors show a legal range of 1.0 (white) to 0.3 (black). The 0.3 thing was really throwing me off, as I thought black was 0.

- RGB 0-255 maps to 1.0-0.3. I displayed two scopes side-by-side to verify this.

- The codec I'm using takes RGB 0-255 and scales it to legal levels (I rendered out a sequence with patches of values like (0,0,0), (1,1,1), (2,2,2), etc. and I could see them all).

What a relief! So, while I wait for the audio line-level converters to arrive, the final step is to draw up some test patterns in Deluxe Paint on the A500 and capture them to make sure everything jives. There might be some codec funny business when I render to MPEG-2, but at least I'll be able to "verify video unity within my editing bay" (that's big boy video professional jargon).

This is awesome. Sometimes I wish I got into video instead of computers. Luckily for me, computers (usually) pays better! -- Trixter

Dec 3, 2006 5:57 pm: Just one little audio problem to fix and then I'm on my way. If you're curious in following along, check here. -- Trixter

Dec 3, 2006 5:57 pm: Happy New Year! 2005 will be a time to shed some pounds and make a DVD. On that note, I just finished getting the A500 set up again and connected to the video rig. This time, the video capture is damn near flawless, but now I have an audio capture issue that I'm trying to resolve with Blackmagic Design's tech support. (It seems that capturing from the system's sound card is a bad idea, since the sound card's clock is different than the capture card's clock. I would very much like to use my system's sound card since it's a great card, but worse comes to worse I can invest in some XLR converters to get around the issue.)

Friends and family may have noticed that, while I am still not reading or responding to email until I get a finalized capture done, I have brought my IM client back up because I'm in such a goddamn fine mood :-) I can be reached via mobygamer @ aim, yahoo, and MSN, as well as ICQ #58657721. Please feel free to drop me a line as I definitely feel like talking about the project again :-)

Life is good! Smile, dammit! :-D

Music: Uncle Tom's music for Mental Hangover (about 20 times), and Demovibes 1 by willbe (which surprised me how well it was mixed -- some of the mixes match up in both tempo AND pitch, which is difficult to do). -- Trixter

Dec 3, 2006 5:57 pm: Movement. I fixed the A500 -- I think -- so I'm ready to start over. Start over? Yes, I found out that there is at least one frame drop in all of the Quicktime captures, so I'm using beta drivers from Black Magic Design to capture native .AVI. A few test captures show no dropped frames, which is good.

You may think that I'm being TOO anal about a single dropped frame. Well, guess again: Not only is it noticable to fanboyz (and myself), but it screws up the "audio conform" process in Premiere Pro and you get stuttering sound 2-3 times in a clip.

Friends and family may also have noticed that I haven't answered email in a week. This is because I am abstaining from anything that distracts me from the project. I will start reading/answering email as soon as I have the first FINAL clip done.

Current music: Travolta's music from State Of The Art. Over and over and over (because that's my test clip right now). And you know what? I never get tired of it! You know you love a piece of music if you can listen to it 100 times without getting sick of it. -- Trixter

Dec 3, 2006 5:57 pm: Holy crap, am I glad I did that! The video rig is TIGHT -- I can capture uncompressed YUV, monitor my capture in Premiere, apply effects and see the changes realtime on the broadcast PAL monitor... I wish I had this kind of quality to work with when doing the first DVD. This is truly kick-ass. As a test, I was able to fix the State Of The Art capture I had previously made. I would still like to re-capture all of my test clips, because I still have some video level unity questions I need to answer for myself, but this is still a great move forward. I feel like I can breathe.

All that being said, I'm taking a break for the holidays, of course. I should be messing around with the project again on the 26th. -- Trixter

Dec 3, 2006 5:57 pm: All this minutia is driving me nuts -- why is the DeckLink capturing fields out of order, why can't I monitor captures in Premiere Pro... so I'm going to completely REFORMAT THE PROJECT:

- All my computers are getting consolidated under the desk. This will let me put more Amigas on the table at one time.

- I am reformatting the capture machine and ripping the Matrox hardware out of it.

- I am reformatting my generic workstation and turning it into a Premiere Pro editing workstation, same as the capture machine.

- I am going to network the two computers together via firewire. This means that, in theory, I can work on either systems' hard drives at drive speed (firewire exceeds the transfer rate of my hard drives).

I need this jolt to get back on track! As it is, I don't have room to set up more than one Amiga at a time, and that's impeding my progress. This will also let me consolidate monitors and I'll be able to replace my editing workstation monitor with my good 19" Sony G400.

Last but not least, this gets all the games off of my machines. No games for me until I finish! Along these lines, I am withholding Half-Life 2 from myself until I finish my portion of the project. -- Trixter

Dec 3, 2006 5:57 pm: Oh ye of little faith: I have indeed been working dilligently on trying to figure out if I have a field order problem in my video setup. I believe I do, so please download this:

ftp://ftp.mindcandydvd.com/pub/MindCandy/Volume_2/tests/TestRender_HalfD1.mpg

...burn it to a PAL DVD, play it on a PAL player and PAL TV, and let me know if the motion is fine or jerky/stuttery. Sorry there's no hyperlink there, I haven't had time to track down what's wrong with my blog code.

Along those lines, do any of our Euro friends know of a PAL DVD that has interlaced video footage on it? I don't care if it's the main feature or just a bonus feature, as long as it's there. I don't want a film DVD, but something with traditional interlaced video on it so that I can use it to calibrate playback on my PAL monitor. Any suggestions?

Music: (None, I've been trying to figure out if/where/whatever my field problem is!) -- Trixter

Dec 3, 2006 5:57 pm: Yes, and now the only blog entry in ten days is about me writing in the blog.

Bad sign, bad sign..

BTW, I encourage everyone who is interested in seeing this DVD completed to email Trixter, and tell him WHY you would like to see it completed. That will help move it along, believe me. -- Phoenix

Dec 3, 2006 5:57 pm: Now, see, that's kick-ass -- that other people in the project are posting to the blog. That was a nice surprise this evening. The 5.1 mixing feen is referring to is one of the nice surprises we're doing for this DVD: a 5.1 remastered mix in addition to the original Amiga stereo hardware output. A third audio track will be commentary by the original group wherever possible.

The capture machine is ready (I still can't preview realtime when capturing to non-quicktime, but I can calibrate it once and forget about it) and I'm clearing off the 240GB backup drives of old junk so they can be ready to... back up the project... Which, come to think of it, aren't going to be enough to back up the project, so they will probably be replaced with two 250GB drives for a total of 360GB capture (RAID 0) and 500GB nearline backup. (Yes, that's a crapload of space. I'm capturing uncompressed YUV @ 8 bits per channel, what do you expect? DV is 3MB/s, my captures are 20MB/s. )

Music: U2's Vertigo (otherwise I dislike U2), Goku, Bowling For Soup, Fountains of Wayne. -- Trixter

Dec 3, 2006 5:57 pm: Pfft.. the first preview clip of Demographics 2 was nothing.. :) the second clip, on the other hand.. much longer, much better. Kids, this featurette is gonna be a whole different animal than the first one.

Now I getta get started working on.. okay, I'm not actually *working* on anything*.. but I'll pick up the music/permission/art coordination efforts as the project grows, mark my word.

* well, maybe the 5.1 soundtracks.. not rushing Jim too much on the capturing though. :)
-- Phoenix

Dec 3, 2006 5:57 pm: A daddy longlegs came crawling toward me from the pile of Amiga Zorro cards I ripped out of the older A4000 when I was troubleshooting it. I blasted it into the twilight zone with a can of compressed air. I have no idea if I obliterated it or if it's quickly scrambling toward another destination -- I can't find it.

I can't tell if this is a good or bad omen.

Music: Taxiride -- Trixter

Dec 3, 2006 5:57 pm: Now we're up to three hours on the computer, at least every other day. Again, nothing directly related to the project, but again this is progress. I am falling back into my old habits, but with modifications to fit my current husband/dad/wage-earner lifestyle. I feel good about this.

All this being said, the new A4000 still sits in the box, mocking me. I am terrified that when I open it up, it will also have limp audio, or a leaking battery, or something else that will depress me. I will not resolve to open it tomorrow, because every time I resolve to do something and then not do it, I get depressed. I can't afford to get depressed again.

Jeremy took a cue from my recent change in behavior and produced an additional minute of "Demographics 2" and, I swear to god, just that one minute really got me pumped -- he has learned Vegas Video to the point of being able to manipulate it like an opera singer manipulates their voice.

Music: The Pursuit Of Happiness (specifically The Downward Road) -- Trixter

Dec 3, 2006 5:57 pm: No progress -- on the DVD, that is. I am here, at the computer, computing away for... (checking the time) about 2 hours now, which is more than I've done at one sitting in the last year. I'm answering email, I'm listening to music... this is progress! Seriously! I used to do this all the time, for 4+ hours stretches; it defined me as a person. It is from that habit and lifestyle that much of my work has come out. I knew things were in trouble when I realized I had a week's worth of email in my inbox. I used to check email ten times a DAY, goddammit. From 1986 to 2002, if you had told me that I would make it a habit to spend less than 2 hours a day using a computer (other than my day job, of course), I wouldn't have believed it. I have to stop this from happening.

So, although no direct work has been done on the project today, it is still progress. And I cleaned up my work area a few days ago, which also helps a great deal (I can set up the Amigas again; previously, I couldn't see desk).

I got mail from Dan Wright today, with a great sticker on the envelope that was, if I'm not mistaken, a screenshot of an Amiga demo. I now feel incredibly guilty that I haven't even opened the other A4000 box yet -- that guilt will probably manifest itself tomorrow since I have to hit the sack (it's 1am right now).

Music I'm currently listening to: Lush.
-- Trixter

Dec 3, 2006 5:57 pm: If anyone had told me that running Amiga demos would be HARDER than running PC demos, I would have laughed at them... and yet I am not laughing. I am running into so many issues. It is rediculous when old finicky demos run just fine on an emulator (buy cloanto!) but wont run on a REAL A500. Its lunacy, I tell you! Lunacy!

First off, early early demos are so tempermental: 512K chipram or 1meg chipram? Kickstart 1.3 or earlier than 1.3? Secondly, early Amigas are PAL and NTSC ONLY -- you can run degrader all you like, but the output display frequency is still not pure NTSC or PAL unless you actually have the right machine for your video standard. I used degrader on my NTSC A500 and, on a monitor at least, everything runs just fine. But when you try to actually capture that 50Hz output into PAL you find out that, oh wait, its actually 50.xHZ where x is something different every time you power the damn thing on. Ville was kind enough to send me one of his PAL A500s... but it just died last week. Ville, Ill contact you later with whats wrong to see if we can fix it.

So, lets just capture everything on an A4000, right? A4000s have a jumper that specifies whether or not theyre PAL or NTSC so that problem is solved... until you capture the audio and realize its screwed up. Two consultations with my DeVry graduate brother and two capacitors later, the sound is still messed up. Its a problem with the -12v power getting to the mini-amp chip, but even though I have schematics to the A4000 motherboard, none of the other components are replacable without true l33t sk1llz so its a lost cause. So I bought another A4000... to the tune of $300 plus shipping. Youd think a 10-yr-old machine would be a little cheaper.

And how does the new A4000 sound? I dont know -- I havent opened the box yet. Im still retrofitting the capture machine. The last DVDs footage was captured with Premiere 5.1c and a Matrox RT2000 over Y/C (s-video) from three different sources (two video cards native output and one crappy scan converter), and as satisfied as I was with the video quality, I wasnt happy with it. This DVD, since were pretty much only getting one shot at it given our previous track record, is getting the most professional treatment we can afford with the profits from the last DVD: RGB output from the Amiga goes into the RGB input of a broadcast-quality scan converter, which converts it to 4:4:4 YUV component output, which then gets captured as 4:2:2 component input by a DeckLink SP card using an 8-bits-per-channel uncompressed codec. Its the shiznit; the conversion is absolutely the very best you can get from a live machine. (Actually, thats not entirely true: I could capture to the 10-bit uncompressed codec, but the datarate of that codec is nearly 30MB/s and the source Amiga, even in AGA HAM modes, doesnt have that kind of color resolution, so its not worth it.) Why tell you all this? Because the capture machine + Decklink SP card + Decklink drivers + Premiere Pro 1.5s scopes/displays arent coming together properly. I cant preview what I capture unless I do it in... quicktime. Blargh.

I expect I'll be talking about the capture board, the computer, our scan converters, editing software, Amiga troubles, and other stuff in the future. Hopefully every single day. Stay tuned. -- Trixter

Dec 3, 2006 5:57 pm: Welcome to the developer's blog, where I plan to talk about my semi-daily progress on the DemoDVD project which, as you can see, has had glacial progress since the release of the first DVD. I expect that only I will be making posts, but others in the project may also be posting (guys, if you need help posting to the blog, let me know and I'll show you the link). I plan to cover mostly technical issues, but other issues will creep in here from time to time; check once a day to see what's happening.

So why has nothing happened for nearly two years? There's no way to sugar-coat it; it's been my fault, solely and completely. I can make excuses, but let's face it: I have a perfectionist personality, and I tend to not want to do anything if I can't do it 100% perfectly. (You can imagine that this tends to cripple my involvement in my various projects.)

I am hoping that daily posts to the blog will help get me motivated (crosses fingers)... -- Trixter