odd stuff: 600MB vs. Direct feeds, docks-scene...

Joined
May 22, 2003
Messages
3,290
Reaction score
1
there's some very fishy about this scene ;)

or atleast I'm like 98% sure that there is...let me explain.

you prolly all think that these two vids are the same, the one just a direct feed vs. shaky cam. right?

well check out the zombie that gets knocked into the water's rag-doll animation...it's different!

in the first(the 600MB shaky cam) the zombiesfalls into the water and seems for a short moment to rise to the surface in a sideways position. in the direct feeds it's stomack down from the start. also the end shot differs...in the direct feed the legs of the zombie is much more spread out than in the shaky cam action. also it's positioning of it's arms seems different(shaky cam they're kind of curled up under the zombie vs. direct feed which got them down along the sides)

it's pretty weird since the rest of the video seems to be pretty much the same except from a quality point of view.

another thing is...am I the only one that thinks that the quality in the docks direct feed video is very poor and rather dissappointing? the lightning and contrast levels seems to be off compared to the shaky cam-vid....funny enough I like the lightning in the shaky version alot better, seems more vibrant...I really hope this isn't just and artifact of the video-cam. also level of detail somehow seems lower...the bridge texture and the zombie texture both looks better(clearer) in the shaky IMO.


anyways back to the rag-doll zombie...why do you think there's this difference... I'm thinking that maybe somehow the engine handles physics in demo-playbacks...like it does in multiplayer...things were exact physics isn't needed might show up different each time you watch the playback(although I'm not really sure if this would be usefull in demo play-back since it essentially makes it real-time) it's the only thing I can think of...since the rest of the vid is exactly the same as far as I can tell. I guess we'll get to know in the other vids if rag-doll effects shows up different.
 
yeah its different each time because dead body physics arent really important to the game so it doesnt save perfectly in the demo :D
 
It doesnt do it per playback , but you have to understand that that presentation was shown many times throughout the day , if you watch the low-quality e3 videos , I have several versions of the same scene. And Gabes text differs slightly , these videos are from different presentations , so this one probably just differs from the e3 video you got , It was from a different time during that day.


Besides the game would have no way to work in a video file.
 
Prince, someone e-mailed either Gabe or Yahn about whether the physics were deterministic (just search the forums for 'deterministic'), he said no, there is a randomization factor, but they could be made to be deterministic through a mod.
 
BTW, gj noticing that. I've been trying to compare the Alyx videos for differences too, but haven't seen anything yet. I think Gabe said that these videos would be different and 'close to final', but I haven't seen any differences at all.
 
Originally posted by DimitriPopov
It doesnt do it per playback , but you have to understand that that presentation was shown many times throughout the day , if you watch the low-quality e3 videos , I have several versions of the same scene. And Gabes text differs slightly , these videos are from different presentations , so this one probably just differs from the e3 video you got , It was from a different time during that day.


Besides the game would have no way to work in a video file.

there were no liveplaying during the public E3 showing(as far as I know)...so it shouldn't matter.

and as I said...the rest of the scene(cam movement etc.) is exactly the same(which would be impossible if it was live playing)
 
Originally posted by dscowboy
Prince, someone e-mailed either Gabe or Yahn about whether the physics were deterministic (just search the forums for 'deterministic'), he said no, there is a randomization factor, but they could be made to be deterministic through a mod.

wasn't that concerning multiplayer not demo-playback though?
 
Originally posted by PriNcE oF SpAcE
there were no liveplaying during the public E3 showing(as far as I know)...so it shouldn't matter.

and as I said...the rest of the scene(cam movement etc.) is exactly the same(which would be impossible if it was live playing)


Well Ill check again and see but Im pretty sure I remember the Watermellon falling different and a few other things.
 
The 600MB video was lower res and slightly blurred and therefore hid a lot of stuff and made it all seem more realistic. Also hid the fact that AA and AF was turned off.

So there are pros and cons, but I the 600MB video actually looks more realistic because it hides stuff... ex. static sky, blurry textures.

Hopefully the static sky and the blurry textures were only placeholders though. I'm not worried about the AA and AF because those were just turned off.
 
MT THOUGHTS

1: My thoughts on the ragdoll physics difference:

If Valve uses a program to do a direct conversion frame for frame from the demo then the first E3 demo was done at a different resolution so it had to be re-converted to this current resolution...therefore the demo ragdoll physics would differ (as it may every time you play a demo)

2: about texture clarity and lighting:

According to Gamespy they did a bunch of post processing of the E3 vid. Not only is it captured with a cam, it is changed by the filters gamespy ran on it...unsuprisingly it is very different.

That completes my 2 cents.
 
Originally posted by dscowboy
Gabe said that these videos would...be 'close to final'...
I don't think he said that at all. These demos are based on an older build of Source. For instance, the skybox in the Docks video is clearly 2D while the recent GameSpy interview said the game will feature 3D skyboxes. These are just demos and not representative of the game's final quality.
 
Erik said they would be close to final aswell. I fail to understand why they would use the old feeds. Why not just run the .dem file now and make an updated feed?

Saying that though i think they did use old feeds cos the crowbar pickup still looks biffy and the placeholders for "green blood" are still there. There is some subtle editing mixed in, like the g-man fading in and out cos theres no gabe to chat in between.
 
Exactly, nothing has changed in the videos. Either they saved an old build from E3 and reran the .dem files, or they made the 'new' movies months ago when they were still on that old build, or nothing significant has changed in the engine since then. I'm willing to believe any of the above.
 
just got reply from Gabe & Yahn....check the Valve info thread...

guess I was right afterall...
 
Originally posted by RUBBERMONKEY
1: My thoughts on the ragdoll physics difference:

If Valve uses a program to do a direct conversion frame for frame from the demo then the first E3 demo was done at a different resolution so it had to be re-converted to this current resolution...therefore the demo ragdoll physics would differ (as it may every time you play a demo)

2: about texture clarity and lighting:

According to Gamespy they did a bunch of post processing of the E3 vid. Not only is it captured with a cam, it is changed by the filters gamespy ran on it...unsuprisingly it is very different.

That completes my 2 cents.

Gamespy put on post processing filters over the original video feed? ... that doesn't sound right to me.

Post-processing filters can be done in real time though. At least it is used in this other game I know...
 
quote

This is copied directly from the gamespy(fileplanet) page:

<quote>
Description
The demonstration of Half-Life 2 was unquestionably one of the hits of this year's E3. All FilePlanet users now have an opportunity to experience this demo thanks to a +500 MB, 25 minute, insanely hi-res QuickTime video.

Update: We increased the quality of this video even further! Enjoy 30 fps of Half-Life 2 joy, plus a brighter picture!
</quote>
 
Well, I think that we will only be able to judge the quality of Half-Life 2's graphics only when we see the new direct feed videos. When I say new I mean "never before seen" new. Also, the game is still in beta form, not final yet. They have plenty of time to fix all the graphical glitches. If I'm not mistaken the original Half-Life had serious bugs and graphical glitches all the way up until the day before it went gold. So it's not final until it goes gold.
 
Did anyone else notice the water horizon? If you look past the sunken ship you can see the water making a square edge where the horizon limit is.. looked a little dodgy I thought
 
That's why I suspect it was just a mock-up level and also explains the mis-aligned textures. After all, the artists and level designers have to get comfortable with the tools before that actually start working on the game, and slapping together a demo level for the press is probably a great way to get started.
 
Plus the build is before 3D skyboxes , dont worry , the water will have a curved edge im sure.
 
I think it's almost the same thing, the shaky cam version looked alot better because the contrast made the shadows and water look a tiny bit better, OR maybe its the dynamic engine scaleing down?
who knows, i think it still looks great :)
 
there were no liveplaying during the public E3 showing(as far as I know)...so it shouldn't matter.

No, the hyrdra sequence was run in live mode.

These are definately from EXACTLY the same build as e3, incorporating NONE of the changes they said they've added since then. My guess is simply that either the latest revision of the engine isn't gold yet for production use, or else they don't want to totally show us everything on this one vid, and basically dont want to show us anything more than we've seen at e3.
 
Back
Top