[Project] G42-200 StarLiner

[Izack]

Some preliminary feedback:
I have a strange phenomenon with this bird, where time seems to progress in jerky steps. The FPS is fine (around 14-18 with a less-than-ideal current processor state) but but all motion seems jumpy. It gets aggravating quickly. Is the G42 doing anything arcane that could cause this?

The three-state switches for landing gear, visor etc. seem like they should only have two states. Flicking up to deploy, then back down, then down, then up to retract is a bit of a pain.

Those are my only two problems. Other than that, on my flight around Florida I fell in love with this bird. It handles beautifully (smoother than the deltaglider and lighter than a space shuttle, but still retaining the feel of being a spacecraft; it's a joy even with the numpad), and looks really nice on the outside, and the cockpit feels like an actual cockpit. The default camera position is excellent, and the three MFDs readily visible without looking down, with one more for anything else the pilot might need, along with the useful info printed to the HUD are very welcome.

This craft has a huge amount of potential. Thank you for that small glimpse in the WIP version.

Seeing as there isn't a reaction control system (although unless you press ctrl+/ it still makes noises like one) I didn't try going into space yet, but I'll go for it tonight. Excellent work, Moach.

[Moach]

most of those 3-way switches will have an automatic pop-back thingy when i get them sorted out...
not sure why you're getting bad FPS... the G42 uses the all the new sketchpad stuff and shouldn't really cause any major FPS hits (ideally)

are you having similar problems with other craft? how less-than-ideal is that processor you mention? i can't guarantee good results below core2 these days...

is your video-card in any better state?

every now and then, orbiter freaks out on me in ways similar to what you said... doesn't seem to be related to any one particular craft... try getting it to enumerate your video devices again...

that got me out of a weird spot last week... (gotta reinstall it, really... but i keep putting it off for i'm too lazy)

[edsupagood]

Quote:

Originally Posted by Izack

 Some preliminary feedback:
I have a strange phenomenon with this bird, where time seems to progress in jerky steps. The FPS is fine (around 14-18 with a less-than-ideal current processor state) but but all motion seems jumpy. It gets aggravating quickly. Is the G42 doing anything arcane that could cause this?

This seems to happen if all the MFD's are on, if I turn them all of the jerking seems to stop.

[Moach]

some MFDs are "heavier" than others... worst of all is the mapMFD, specially if you have it on "groundtrack" mode... HSI and surface are also heavies

you can just turn them off when not using... seems to help in not-so-recent CPUs...

not much i can do about that, unfortunately... that's mostly a Martin issue



if someone could provide feedback on how it runs with non-stock graphical clients... that'd be nice too

[Grover]

ok, my own initial feedback:

same FPS problem as Izack, and my computer is <1 year old, and can have a full energia stack, with BuranT, two tugs, and two vessels, as well as run orbiter with space stations, somethings fishy

the RCS is overpowered, and the OMS is underpowered, i think reducing the RCS by 30% and increasing the OMS to 3-4 times its current power, it looks like im getting 0.5M/S^2 at the moment

MFDs still cannot read the thrust of the OMS engines correctly, IMFD wont even run autoburns because it thinks there isnt an engine there (coz there isnt as far as it can tell), at least the OMS should be given top priority for Main engine authority, then you could use the Main/OMS switch to kill the OMS engine anyway.

without cockpit illumination, i suffer from the same problem as with the '100 WIP: i cant see around the cockpit for half my orbital time (yes, i know its on a later WIP)

my final criticism is that the trim is more effective than the main controls :dunno: and the control selectors for Airfoils and RCS arent working yet


besides the criticisms, here are my happy-Bonuses :D:D

i got it into orbit on my first attmept with plenty fuel for de-orbiting

the VC is complex, but easy to understand, once you know where the key switches are located

the feeling after seeing the visor roll down, to reveal the glorious Earth made me weep inside


thats all i got for now, ill give it another go, but those are my primary concerns for the moment, lets see if i cant make that launch a little more efficient :D

---------- Post added at 07:54 PM ---------- Previous post was at 07:53 PM ----------

EDIT: ill try turning off some MFDs, that might help the lag issue

[edsupagood]

On D3D9Client I'm getting over 150 FPS exterior, but in VC mode FPS goes down to 5 - 10. Don't know what is causing it, turning of all MFD's increases about 5 FPS.

[Izack]

Fair enough for the switches. That's even better.

I'm on a laptop, which has a 2-core 2.1GHz processor. Video card is...yeah, let's not go there.

I've tracked the time problem down the the G42 virtual cockpit. When flying any other vessel, even in the same scenario right next to the G42, and on the G42 itself in glass cockpit mode, the scenario runs smoothly. Load the VC, and time starts hiccuping, in internal and external views. Switching to another vessel or pressing F8 to change to glasspit returns the simulation to normal.

[Hlynkacg]

Just downloaded it but I'm in school right now and wont get to really fly it till this evening. I'm looking forward to it.

[Grover]

my own experiments found that disabling MFDs did almost nothing, but going to timewarp 0.1 improved my FPS to nearly 60. im not sure whats causing the problem, but from what i read, its related to the VC, has wide impacts and affects every computer out there. whilst it is bearable, its annoying nontheless, perhaps we can work out whats causing it and get back to our 30-40 FPS guys?

and ive only just noticed how powerful those engines are, pitching to 40* and still accelerating? hell yea!

im gonna try a flight, KSC>WIN at just above concorde levels, see how it can handle in the atmosphere for now, then ill see how we do in space once we work our way into further WIP releases, ill definatley get excited and take it to space anyway, but ill try and keep to the atmosphere, coz someone's gotta do it

[Moach]

ok, actual bugs bugs so far:

• overpowerd RCS
• underpowered OMS
• trim more effective than ACS
• bad FPS in VC

as for the lights thing, as well as the IMFD-cant-read-OMS-thrust issue, those are "WIP-faults" and should be ignored for now, if at all possible


i'm almost certain the FPS issues on the VC are due to the oft-redrawn EICAS page... unfortunately, splitting it into separate textures will only make things worse, and i can't really make it a smaller texture, or it'd get so pixelated you wouldn't be able to read from it....
i could have it update in sections, such as only one "screen" per frame, but that would make for a possibly clunky refresh rate



so now i'm gauging the possibility of a DirectDraw-based implementation... which would divert the redraw load to the GPU and free you poor CPUs of that perhaps abusive burden

that would not be standards-compliant, tho - the current implementation uses the oapi::SketchPad system, as by "Doctor's orders", using DirectDraw would pretty much be like walking all over that and bypass most of Orbiter's 2d-rendering engine...

i'm sure Martin didn't intend this to be the "proper way" of doing things when he brought us the SketchPad... and i can't be sure on how that would sit with other graphics clients (specially OGLA)


i think it's possible to optimize the redraw by breaking it down... not sure how orbiter would like it if i fired up another thread to draw on it's textures... that's another possibility to investigate

Grovers mentioning that turning the sim rate to .1 boosts FPS is a major tell-tale sign for me... for i manually command a redraw of the EICAS page every 120 milliseconds (of sim-time, which was probably a crappy idea now that i think of it )

so that's a good pointer to where i should look for trouble...

i'll se what i can do to get better FPS around that

[Grover]

it seems Moach has the brain of Hawking!

ive had trouble replicating the part where going back to .1 timewarp was helping, but i also failed to replicate engine startup for a few minutes, so i guess im just tired

and ive found that the main engines (no AB, no RAMCASTER) can reach around mach 5 (at around 20k alt mind you) so the main engines arent as underpowered as the readme (and yourself) made out to be, perhaps you could add a multiplier to the Altitude (or static pressure) effect on main engine performance, because i doubt any turbofan engines can either A) get that high B) go that fast.

the whole craft also seems to have developed a tendancy to roll left. if left to its own devices (whilst i was at 20k with my mains) the craft simply wouldnt hold its attitude at all, even the WINGLVL AP had some trouble. ill reset my joystick and try again, dont take my word for it just yet, perhaps it is linked to the "Mesh migration" you noticed a while back? when working on the VC

also, some cosmetic bugs (texturing may remove some of these bugs)

the two piece visor doesnt fit perfectly, you can see the black-drawn area beneath them

some of the vertex normals around the flanks of the RAMCASTER (near the engines) have gone AWOL, leaving an arkward reflection on them (ill try and get a screenie)

whilst the FPS is low and switching seats (which takes a moment too long for now) you can see the pilot meshes jerking as new ones are loaded

the light switches' labels are hard to read due to their guards, perhaps you could make then a little smaller so the labels can be read more easily

the current text on the EICAS is a little too small, i have decided on 40* FOV so i can see enough of the cockpit at once, and the text remains too small to be read, even on 1600x900 display



thats enough for the bugs for now, i guess id better have another go at getting around KSC (a few flights of the pattern seem a good idea)

later dawg!

[jangofett287]

I've noticed a tendency to roll left as well, and I'm using numpad, no joystick at all.

[Moach]

Quote:

Originally Posted by Grover

 it seems Moach has the brain of Hawking!
(...)

not even close! (i only wish)

if you see that it takes time for switching between the two pit-side meshes, then your pc is probably well beyond what could be considered "recommended specs"... sad but true...


my laptop suffers quite a bit to run 3d stuff as well... i don't think 3d is its strong suit


as for the screens being too small to read, i'm aware of that - they're also excessively high on resolution... next version will get a remapped EICAS page, so that each display mode is only drawn once, regardless of which screens are set to render it (should use UV mapping for that)

and the resolution should be about .75 of the current for each screen... which automatically would make the digits larger and easier to read


anyways, i might be able to completely remove the stuttering if i draw the EICAS page in a separate thread, then oapiBlt it into the surface when a full oass is complete (which apparently is done by GPU, thus: fast)

a mutex would hold the redraw thread until all the blitting is done and new flight data is ready to be drawn, then the cycle restarts

it's a rather nice solution, i believe... it should completely remove any stutters, since most CPUs nowadays are 2-core (or more) and multithreading is becoming the standard way to deal with stuff like this....

still more braining to do on this...

but now, back to work... ("work" work, that is... no Orbiter work right now)



when i get home maybe...

[Grover]

just two little things for now:

the nosewheel steering is way hyperactive, it needs to be toned WAY down somehow

when ram-starting the mains, its hard to tell when they're going, we need an easy-access display to tell us when theyre running, throttle setting (already on the HUD), and actual working throttle (these are the minimum IMHO) perhaps putting a Boeing-style engine Revmeter (or whatever theyre called) on the EICAS?

so, to summarise all the bugs found so far (to make it easy for you Moach )
RCS overpowered
OMS underpowered (you'll need 1.5Mn total thrust to give 7m/s^2 acceleration (is this enough) for an optimistic fuel load (if you did a really efficient launch)
the FPS problem
engine definitions not available to MFDs
trim too effective
Main engines effective at too high airspeeds and too high altitude (if you can fix this without affecting the speed/alt for transitioning from main/RAMCASTER-LOW i will be incredibly impressed with your skill )
the two pieces of the visor dont seem to fit perfectly (textures may fix)
vertex normals on the edges of the Gear Door dont match with the adjacant hull
VC light switch labels are hidden by the switch guards
small EICAS text
constant left roll, most prominent with wings down and at high speed
Overactive Nosewheel steering

and thats all for now

i gotta say, its a shorter list than some of the other BETA tests ive done, and ive been a little critical myself :sorry:, but there doesnt seem to be a huge amount of required work, and for a WIP-1, this is awesome stuff!

later dawg!

[Eli13]

Huh, I downloaded everything correctly and everything is in the right file and I even used a clean orbiter slate, but.... when i start the scenario it immediately crashes. I haven't even seen the G42 yet. Except on here of course.