ahh okay. Probably could have found that out with a little googling. I hate my laziness. Anyways, hows the run going?
We’re pretty sure we’ve found everything in Chapter 1, we’ve just begun recording segments.
The idea at the moment is to examine a chapter, run it, examine, run it etc.
Progress is being recorded on the timesheet I posted above.
Just a thought I had after reading your post.
Instead of examining new chapters only after you finish one, you should plan a rough route for the rest of the game so you’ll know how much health you’ll need in the future chapters and whether you are able to squeeze in a damage boost (or something that requires health) or not.
I like to put up lists with important events to make it easy and lucid.
The HL21 plan sheet got 233 lines/events for comparison : >
Well… looks like I’ve discovered a major issue.
When comparing Takyon’s segment demos to mine, I noticed that the airtime for the jump intervals with my demos are a lot slower than Takyon’s, leading his segments to be significantly faster and way more optimised than mine.
We both examined each segment with viewdemo and noticed that the airtime for my jumps is roughly .70, whereas Takyon’s is .66
When I first started out running Gunman I noticed that the jumps seemed a lot floatier than in HL, at first I thought it was to do with slightly altered movement mechanics in Gunman, but reviewing Takyon’s demos this doesn’t seem to be the case.
First of all we thought it could be the script I’m using, so I tried out the one that Takyon uses (which is exactly the same apart from it having one less wait), there was no difference.
I then noticed that despite me having fps_max set to 250, whilst bhopping my FPS dropped to roughly 200fps.
I made a comparison video, watch the video below and see what I mean, I’ve also included the demos in a downloadable ZIP file:
Comparison video (apologies for the bad quality and frame drops, my computer sucks with Fraps):
As you can see in the comparison video, my jumps in airtime are a lot slower than Takyon’s. In the second part of the video, I used host_framerate to slow down the demo to 1000fps, if you look closely, you can see that my demo shows that I get increases of speed in small chunks, whereas in Takyon’s it is completely smooth.
Obviously this needs to be solved ASAP as I am currently at a rather big disadvantage when it comes to optimising segments.
UPDATE: Here’s the current config I’m using, just in case it has anything to do with it:
alias +aut “alias name bhop;bhop”
alias -aut “alias name”
alias bhop “+jump;wait;-jump;cmd name”
bind SPACE +aut
//replace on pause0 line to your sensitivity value
alias pause1 “host_framerate 0.000000001;sensitivity 0;bind mouse2 pause0”
alias pause0 “host_framerate 0;sensitivity 1.35;bind mouse2 +attack2” //SENS&BIND
alias stopp “fps_max 250;pausable 1;sensitivity 0;pause;w4;stop;pause1;bind mouse2 pause0”
bind TAB stopp
bind 4 pause0[/code]
UPDATE 2: I definitely do not think it has anything to do with FPS, as I’ve just tried changing to 100 and 60 and it hasn’t made a difference.
I suspect your unstable frame rate anyway. See link to see how much fps can affect the player speed.
Also host_framerate speed depends on fps. If a lot of gibs spawn for example you can clearly see a decrease in rendering speed.
I suppose you use the older version of NGHL? What you can try is to install the latest one with the fps fix applied (fps doesn’t change physics).
Another thing to try is to record a demo while r_norefresh 1.
Problem solved, changed fps_max to 125.
I have the latest version of NGHL; fps_physics_fix is set to 1 by default, although that command isn’t featured in Gunman, so I’m not sure it would help anyway.
Recorded a demo with r_norefresh 1, watched it, still had frame drops when fps_max was set to 250.
EDIT: Out of curiosity, I watched quad’s bunnyrace video then compared it to one of my demos, and my jumps seem slower there too, so it’s possible having frame loss has actually had a negative effect on all of my current runs, although I’m not even sure since fps_physics_fix is set to 1 like I said.
Although it’s an engine fix, it’s done through memory editing it from client.dll (RCD) and not some extra DLL.
So can RCD be used to apply the fix to your bunnymod also?
When altering my fps_max to 125, I have bad timing when bhopping, of course when it’s set to 250 the timing is fine but obviously the framerate becomes unstable again.
Unfortunately it cannot, and I don’t plan to implement that fix into Bunnymod Pro, because if launched under Steam it can lead to a VAC ban. If you’re in a dire need, you can install Bunnymod Pro into the latest Steam Half-Life which has the fix built into it.
I made this thing: https://www.youtube.com/watch?v=5vNimBzEHhk
Kind of escalated into something else but I guess that it shows that we’re getting somewhere and have started recording segments.
Yeah, I guess I should elaborate on that…
I’ve had a lot of people asking me what the progress is like, some even asking when it’s going to be done, so I’ll talk a little about the progress without spoiling too much because I honestly believe that this run will be worth the wait.
I think the first thing that needs to be considered is that we’re trying to make this as optimized as possible, a Gunman equivalent to HL21 if you will, but with only one previous run being made beforehand, which we can barely even take reference from anyway because of how unoptimized it is. Saying that, considering how long ago this thread got made, the amount of stuff that got found in the meantime is large and when combined with the stuff that me, Takyon and some others found after starting up the project, I think it’s safe to say that the majority; if not all of the major skips/tricks/glitches have been discovered. At this point we’re focusing on the more detailed stuff and I’ll say that a little under a quarter of the game has had a rough route drafted out for it.
What excites me most about the routing in this run is most chapters each have a specific trick that gets abused the most, for example in the Mayan levels of the game (the second chapter) the route will mostly consist of constant object boosting. Here’s a screenshot of the route folder for Mayan that includes every single object boost that can be used, as well as some other cool tricks/skips:
Segment-wise, we have only really just begun. Of course some will know that I’ve had some technical difficulties based on framerate and I hope to fix that as soon as I can, but that unfortunately is the sole reason why we haven’t really gotten much done in that area. Besides, I personally think that the rough planning and routing is more important at the moment.
Anyway, to conclude I guess the point is that whilst you might not have heard much and no matter how slow the progress is, the run is far from dead and it’s still the project I’m looking most forward to completing, as well as it being my highest priority speedrunning-wise ;D .
More interesting stuff will be revealed as stuff actually gets done, I hope that when we have a decent amount of segments done that we can get a trailer rolling to hype up the run before it’s release
Oh and if you have any questions about the run, don’t hesitate to ask!
Did a bunch of weapon boost tests on bhop.bsp : https://www.youtube.com/watch?v=b1vJav_PIBU
Found a new technique of switching weapon back and forth to cut out some waiting between shots. Usual HL-style fastfire by save/load doesen’t seem to work in Gunman however.
The switch technique have been mentioned by TYPICAL before. Nice sg+chem combo though. But that might be hard to perfect while bhoping for real.
I see, it only seems to work in different combinations that involves the shotgun (though I haven’t tried all weapon combinations). Yes, it’s tricky to fit it in a loop script in a good way because of all the waits needed for perfect weapon switches. This is how I did the perfect timing for the video:
Host_framerate 0.001 // fps_max 1000
lastinv // shotgun
lastinv // chemgun
lastinv // shotgun
lastinv // chemgun
lastinv // shotgun
lastinv // chemgun
Will likely only work as a quick initial boost in the beginning of hallways etc. in practice.
Edit: Also I only seem to be able to use lastinv instead of just weapon_chemgun/weapon_shotgun or invprev/invnext for this. Not idea why.
So i tried out this game on my stream, and PJC was in skype and said that that a feasable RTA skip of that section could be useful, but not yet really discovered. So i came up with this skip. I don’t know if its useful or saves any time, but yeah.
So what’s the status for this run? It says it’s coming soon in the header!
if im correct, coming soon is in valve time
exec wanted to make a banner for it which is why it’s there, however I’d imagine it won’t be here for another year or so.