Strange issue with Phantom...

"How To" by our Pilot Staff
Post Reply
User avatar
Teej
Virtual Thunderbird
Posts: 1533
Joined: Fri Jul 31, 2009 9:29 pm
Location: Milwaukee, WI

Strange issue with Phantom...

Post by Teej » Fri Aug 07, 2009 2:15 pm

OK...I'm working from track files that I am quite sure are valid track files.

I have followed the instructions accurately. On two different machines that couldn't be less alike.

1:
Dell Vostro 1510 Laptop
C2Duo T5670 @ 1.8 GHz proc
2 GB ram
Intel graphics
Vista Home Basic

2:
Self-built desktop
AMD 64 X2 4400
2GB
Nvidia 8600GT graphics
Fresh (less than 24 hours old) install of XP SP3

Neither has any sort of input device that would've been activated while running Phantom.

In both cases, they run just a _little_ bit slower when playing back as a phantom...pretty much imperceptible while flying except when you see the cumulative difference in position which results in the lead aircraft bouncing off the dirt after about 10 minutes.

It's worth noting that with either machine...lead bounces at _exactly_ the same spot. Assuredly does not happen in the source track.

Any thoughts?
User avatar
Thumper
Virtual Thunderbird Alumnus
Posts: 833
Joined: Thu Nov 03, 2005 10:00 am
Location: Weslaco, TX

Post by Thumper » Fri Aug 07, 2009 2:29 pm

You might try re-tracing the track, if you haven't already. It's possible that you got a bad trace for some reason. Make sure when tracing that you don't have any extraneous processes/programs running and that you don't interact at all with the trace machine once tracing begins.
Image
User avatar
Lawndart
Virtual Thunderbird
Posts: 9290
Joined: Tue Mar 29, 2005 8:36 am
Location: Mooresville, NC

Post by Lawndart » Fri Aug 07, 2009 2:43 pm

Phantom is really finicky.

When recording tracks for phantom use there can be no harsh inputs and the pulls need to be very steady (even if hard). Unloads for example do not trace well and you'll end up having the scenario you're describing with the jet straying "off course" and eventually hitting the ground. Sometimes it takes several tries to create a good enough track for phantom flying, and the tracks we use on the VTB have generally passed our QA to work from start to finish. I'm afraid you'll have to go through this painstaking testing with each phantom track you create...

A few pointers: Remember to always raise your altitude parameters for Phantom recording! This in itself is not the cause nor the fix for Phantom straying off course in a trace, but since you always have more time and room to fix any hairy pull smoothly and deliberately when you have more altitude, that is definitely one deliberate change you should make for Phantom tracks.

Remember that Phantom doesn't know your altitude or how you are flying (upside down/right side up etc), it simply records your x/y-axis inputs and over time any "abrubt" inputs or "spikes" in the x/y-graph seems to be what takes the track off its indented course (compared to the way you saw it while flying it). This is possibly because phantom tracing isn't as precise in polling your stick inputs as your Cougar/stick is. At least that's my theory and the remedy is to be even smoother than normal at all times, plus give yourself higher mins so you have a buffer to always ease off/add more G smoothly despite if you botched maneuver. Phantom doesn't take well to quick onsets of stick inputs and/or a fast correction in an opposite direction (such as an unload). One area of particular concern is the bottom of any Split S or Half Cuban, the G offset needs to be constantly and smoothly released. The added altitude gives you time and space to be smooth despite maneuver errors...

Think of Phantom trace as a slow end computer with bad frame rates that looks at each frame of where your x/y-axis is and takes a snapshot, and then interpolates between the frames to recreate your stick movement. Then think of your actual flying (recording it) with good frame rates and that you may be able to "see" more than phantom trace does. In other words, if you unloaded fast it may have been smooth on your end, but the way phantom "sees" this when polling the x/y-graph because of its refresh rate could mean it "missed" a number of small inputs and had to interpolate your pull between fewer graph points, thus making an unload or pull more abrubt than it really was recording it. I'd recommend two rules for phantom recording:
  1. Fly every maneuver 150' AGL higher than its target alitutde where it would normally be flown.
  2. Always use constant onsets/offsets in pull (i.e. be smooth), even for good pulls exceeding 3+ G's. Think of the framerate analogy, and how it needs to look smooth even on a low end system (that's essentially phantom trace vs. your Cougar/stick).
Aside from that, I'd second what Thumper said about processes running in the background or anything you may have done during the tracing to interrupt it. Be sure not to touch your computer, period - when tracing a track and save the .phtrk immediately once done.
User avatar
Teej
Virtual Thunderbird
Posts: 1533
Joined: Fri Jul 31, 2009 9:29 pm
Location: Milwaukee, WI

Post by Teej » Fri Aug 07, 2009 4:51 pm

Well...from some other testing it looks like the problem is not with the trace process. It's all in the playback.

It's like Phantom is running just a fraction of a second too fast. I did think of one possible thing, although I find it hard to believe this would do it...

Because my laptop is a bit slower, I figured I wouldn't load it down with Ventrilo/voice. I was playing the voice back on the machine I was actually flying from.

Think the fact that it is NOT running through the Ventrilo playback has anything to do with it?! (Again, I'd find this hard to buy, myself...)
User avatar
Lawndart
Virtual Thunderbird
Posts: 9290
Joined: Tue Mar 29, 2005 8:36 am
Location: Mooresville, NC

Post by Lawndart » Fri Aug 07, 2009 6:04 pm

I don't think it has anything to do with Ventrilo, but you can troubleshoot (and narrow down the problem) by not even running Ventrilo at all and play the track and see if it does the same thing?

Secondly, try running the .phtrk without anything else running in the background, aside from what's necessary for Windows and graphics drivers etc. and see about the outcome.

To further more narrow down possible causes, are you absolutely sure you've followed all the instructions in Creating Phantom Control Tracks to the tee? I don't mean to imply that you overlooked something, but 9 times out of 10 that's exactly what happens when people have problems with Phantom Control. :wink:
User avatar
Teej
Virtual Thunderbird
Posts: 1533
Joined: Fri Jul 31, 2009 9:29 pm
Location: Milwaukee, WI

Post by Teej » Fri Aug 07, 2009 6:10 pm

Lawndart wrote:I don't think it has anything to do with Ventrilo, but you can troubleshoot (and narrow down the problem) by not even running Ventrilo at all and play the track and see if it does the same thing?

Secondly, try running the .phtrk without anything else running in the background, aside from what's necessary for Windows and graphics drivers etc. and see about the outcome.

To further more narrow down possible causes, are you absolutely sure you've followed all the instructions in Creating Phantom Control Tracks to the tee? I don't mean to imply that you overlooked something, but 9 times out of 10 that's exactly what happens when people have problems with Phantom Control. :wink:
re: Ventrilo - that's what I was saying - neither machine I was attempting to use as a phantom server had Ventrilo on it. Both machines the phantom inputs seemed to be getting slightly ahead of lead's jet. I was throwing out the possiblity (not that I considered it a legitimate possibility but it's the only thing I'm aware of that's different between us) that because Ventrilo wasn't there chewing a few cycles and possibly buffering some inputs...maybe that was the difference.

I highly doubt that though, given that many (most?) people probably run phantom just fine w/o Ventrilo...


re: Creating the tracks - per our discussion, I think we can completely rule out a mistake on my part during the track creation given the result of that last test (lead jet hits exactly, precisely, down to the foot the same as my previous attempts.) And it happens whether it's on a laptop with default lomac graphics settings....same laptop with everything "minned out" in modman....and a freshly installed desktop with nothing but lomac on it.

Very strange.
User avatar
Redeye
Virtual Thunderbird
Posts: 852
Joined: Wed Mar 30, 2005 1:48 pm
Location: Vancouver, BC

Post by Redeye » Fri Aug 07, 2009 8:02 pm

There has been trace problems when a joystick or controller is installed on the same machine you want to make your phantom server.
User avatar
Lawndart
Virtual Thunderbird
Posts: 9290
Joined: Tue Mar 29, 2005 8:36 am
Location: Mooresville, NC

Post by Lawndart » Fri Aug 07, 2009 8:15 pm

Thank Thumper for this tip: If the track was recorded using "unlimited fuel", be sure that the server runs this setting during trace and playback as well (otherwise the slight weight shift would cause the x/y inputs to affect the jet differently as fuel is burned off if the track was originally recorded with a constant/unlimited fuel load).

After Thumper mentioned this, I definitely agree with him that this could be the culprit! Change your options to unlimited fuel.
Last edited by Lawndart on Fri Aug 07, 2009 9:33 pm, edited 2 times in total.
User avatar
Panther
Virtual Thunderbird
Posts: 1009
Joined: Wed Sep 13, 2006 4:41 am
Location: Las Vegas, NV

Post by Panther » Fri Aug 07, 2009 8:26 pm

I had similar issues with Phantom when I started using it. One thing I did that was causing the aircraft to hit the ground continuously, was using "Show Replay" to trace the track. I didn't realize to trace the track, it has to be played via "Open Mission" drop down box on lower right "Tracks" and double clicked the file and then Click Play track or "ctrl +p".

Just my .02
User avatar
Teej
Virtual Thunderbird
Posts: 1533
Joined: Fri Jul 31, 2009 9:29 pm
Location: Milwaukee, WI

Post by Teej » Sat Aug 08, 2009 12:00 am

Lawndart wrote:Thank Thumper for this tip: If the track was recorded using "unlimited fuel", be sure that the server runs this setting during trace and playback with the track as well (otherwise the slight weight shift would cause the x/y inputs to affect the jet differently as fuel is burned off if the track was originally recorded with a constant/unlimited fuel load).
Probably the most likely candidate so far...definitely worth a try later. Don't have time at the moment.

@ Red: Neither computer has ever had a joystick...or anything else besides mouse/keyboard. No TIR, no Cougar, nothing.

@ Panther: Nice thought. I had guessed (and had confirmed) that the play track was necessary...that's not my issue.

I've got hopes on the unlimited fuel suggestion...


OK, I wrote the above quite a bit earlier. I thought I submitted it. I guess I hadn't.

In any case...the answer suggested by Thumper through LD...proved to be the case as I suspected when I read it. _EXCELLENT_ call.

/me grabs the Jeremiah Weed out of the freezer and pours a round. ;)
User avatar
Thumper
Virtual Thunderbird Alumnus
Posts: 833
Joined: Thu Nov 03, 2005 10:00 am
Location: Weslaco, TX

Post by Thumper » Sat Aug 08, 2009 12:05 am

WOOHOO!! I had the same issue when I started running a phantom server two or three years ago. Server settings... As I said to LD earlier, I take the server settings for granted so much these days that I forget to mention them when troubleshooting issues but the "unlimited fuel" issue will get you EVERY time.

So glad you got it sorted out, Teej!
Image
User avatar
Frazer
Posts: 507
Joined: Tue Nov 15, 2005 8:47 am
Location: The Netherlands
Contact:

Post by Frazer » Sat Aug 08, 2009 4:08 am

As an early Alpha tester of PhCtr, I can add one possible issue to look at.
Make sure that (Lock On\Config\Joyrange) is at its standard value "JoyRange = 400;". We found out that the chance of having a faulty record is reduced by doing that.

We flew solo aerobatics tracks with good results aswell.

Please note that usually only 1 out of 3 flown tracks work as they should. There is almost always a slight shift between the flown track and the traced .phtrk. The easiest way to make this not noticable is to start and end the mission in the air. :wink:

More Info:
Phantom Release post

Phantom Release post (RU)

Berkuts Phantom Server
Post Reply