Been a while, I keep remembering that it's been a decade since I started console scripting in source games. Thought I'd drop this quickie announcement in.

Ten years is a long while. When I first started in HL2DM I made basic scripts without many advancements for 2 years, after which I moved on to TF2 and tried to bypass ent_fire through the use of outputs on different entities. That's when I discovered logic entities, and the whole IO system used for mapping. Obviously I took off, I just didn't expect to be so far ahead.

Seeing people talk about me is weird. I'm old and tired and I don't have any more energy left, really. VScript is obviously the next step, but these days I prefer to script alone, away from public sv_cheats servers. Guess we'll see how far I go with that.

Here's to another 10, yeah?

... idk lol

I lived. Sorry for the anticlimactic announcement, but I'm back, for now.
The previous announcements will be nuked soon.

thx for the help, alex

Yesterday I released an item to the Steam workshop for TF2 that plays Bad Apple!! at a stable and somewhat acceptable quality. On a 32x24 resolution, using box models, at 20 frames per second, and at a size of 108 MB uncompressed.
https://psteamcommunity.yuanyoumao.com/sharedfiles/filedetails/?id=2881236031

The whole idea of playing Bad Apple!! in Source fiasco began about a year ago, when I first discovered videos that do the same thing but on different outputs (Google Maps, Graphing Calculators, Task Manager with a 896-Core Processor... just to name a few).

I was mesmerised and wanted to try doing the same thing myself. Out of all options available I decided to achieve this on HL2 props in TF2. But of course everything went downhill from there. I didn't have enough programming skills at that time to even get simple image data from a video. And remaking the video manually would take longer than the Universe can last.

Fast forward to sometime on 20th of May, I actually had a proof-of-concept way of achieving this, and started work. In just 2 days, I achieved it, and it ran on console scripts. Then this video surfaced. It wasn't too popular for my taste, but I ain't complaining.
https://www.youtube.com/watch?v=eqRL2tPP85g

The map idea stemmed from the video posted above. In the video I strung together a bunch of console commands in over 2000 .cfg files for updating frames of the video, 10 times per second.

At first I was proud of it, but the more I watched it I kinda hated it for being too offsync, all because of a fetid command called "wait". It delays consequent commands in the given amount of frames. Add that to the already constantly fluctuating FPS of the game and the problem becomes clear now. I wanted to use a consistent method of delaying frame updates.

So at the evening of 27th of October this year I began work. I knew that if I was gonna try and make it consistent I would have to make it a map. I decided I would try and make it run on 25 frames per second. 2.5x more than last time!


... I asked myself that every time I got into roadblocks. The first one being the edict limit of 2048.

To put it simply: if I add and surpass this many entities into a map, the game instantly crashes. There's no warning in the console, in-game or anywhere. It just errors out. Thanks Ed Alloc.
Thankfully I narrowed it down using cases. So instead of there being more entities than the current map, there was now about 1200. Still plenty of room for the player to get in and experience the map.

Second roadblock came in rather late when I initially finished rendering the video: the hard limit of entity info a map can have.
I was stuck here for longer than I expected. I thought I could just fix it by optimizing outputs, i.e: target all pixels if they're all the same color on a frame, but no. This time it was harsh. I narrowed down the color palette to only 5 colors, and that didn't help either. So I binned the idea for a bit, then came back to it after a good night's sleep.
The moment I started work on the map again I decided to lower the framerate of the video to 20 frames per second, while also keeping the optimizations I added in the first place. I rendered it and went out for some dinner. Lo and behold, the engine liked it, and after loading into a map, it worked.


Initially, the map was meant to only have the HL2 crate model as the pixel for every prop. This all changed when I started testing out the map with various settings and on various computers. My friends always complained:

引用自 a few friendly folks：

HL2 crates suck. You should make it run on fullbright white boxes.

I didn't really like the idea of using white boxes initially, but after being convinced hard enough, I decided to do it. So I whipped out a quick 40x40x40 cube model and added an option in the map to switch between it and the HL2 crates. The said friends were happy about this change, and I was too.


This map was... something to work on. And even though it's not groundbreaking, I (yes, myself) think it's a cool achievement nonetheless. Still unfortunate that I had to lower the FPS from 25 to 20, but sacrifices had to be made I guess.

But what's in store for the future? Well I certainly don't wanna work on maps like these ever again, so please don't suggest me making Lagtrain in TF2 next time. You won't be able to convince me anyway.

See you next time when I make tetris in TF2!

i'll drop in an announcement to wake all of you up, sorry

speaking of which, should i even post announcements and events? are you all going to be annoyed by them? please do tell.

this group is now public because i found out that i'm too lazy to check requests, and so most of you were stranded in the awaiting for approval status

so now, not anymore