New things from Soji

News, announcements, updates, & more regarding YSFlight & YSFHQ
V8-DLA
Contributor
Contributor
Posts: 434
Joined: Mon Aug 10, 2015 1:35 pm
Favorite Aircraft: Boeing 787
OS: Mac OS X Yosemite
Has thanked: 111 times
Been thanked: 52 times

Re: New things from Soji

Post by V8-DLA » Sun May 22, 2016 12:30 pm

NajmiCreative wrote:recent post from Soji...just take the last paragraph and left all of it
Soji wrote:I need to resume YSFLIGHT development as well. .... Where was I stopping? I feel like decades have passed since I saw the YSFLIGHT code last time.
at last,the development continues back... especially for mobile ones (I guess? :? )
clever find there,hope he develop the mobile version that might soon be released,might try to install addons if it's finished.
Currently inactive,may return soon.

User avatar
u2fly
Veteran
Veteran
Posts: 1191
Joined: Fri Oct 02, 2015 9:27 am
Favorite Aircraft: Antonov An-225 Mriya
Location: Ukraine
OS: Linux
Has thanked: 669 times
Been thanked: 267 times
Contact:

Re: New things from Soji

Post by u2fly » Sun May 22, 2016 2:10 pm

NajmiCreative wrote:recent post from Soji...just take the last paragraph and left all of it
Soji wrote:I need to resume YSFLIGHT development as well. .... Where was I stopping? I feel like decades have passed since I saw the YSFLIGHT code last time.
at last,the development continues back... especially for mobile ones (I guess? :? )
Thanks, good news.
/!\ READ YSFLIGHT HANDBOOK (.PDF) | updated 2017/05/03

U.S. Senator John McCain wrote:
“Spending #NewYearsEve w/ brave #Ukrainian Marines at a forward combat outpost - we stand w/ them in their fight against #Putin's aggression”
http://pic.twitter.com/vkz0gdozVV
(31 DEC 2016)

User avatar
Gunny
Business Class Membership
Business Class Membership
Posts: 3195
Joined: Sun Feb 06, 2011 6:17 pm
Favorite Aircraft: Any thing with a prop.
Location: New Iberia.LA.
OS: Windows XP SP3
Has thanked: 545 times
Been thanked: 690 times

Re: New things from Soji

Post by Gunny » Sun May 22, 2016 8:01 pm

Truly frustrating to me that Soji has not made any progress on the main program but yet detours to develop mobile apps :(
Image[/align=centre]

User avatar
u2fly
Veteran
Veteran
Posts: 1191
Joined: Fri Oct 02, 2015 9:27 am
Favorite Aircraft: Antonov An-225 Mriya
Location: Ukraine
OS: Linux
Has thanked: 669 times
Been thanked: 267 times
Contact:

Re: New things from Soji

Post by u2fly » Thu Jun 02, 2016 2:59 am

Few month ago I wrote a letter to Soji with information about PolygonCrest problems and now got an answer:
Soji Yamakawa wrote:2.06.2016, 03:29
Re: PolygonCrest

Hi U2fly,

Thank you for the E-Mail and I'm sorry for late reply.

I'll soon upload the new version of Polygon Crest. I was extremely busy for the past six months or so, and my out-of-work activities were very much reduced. But, I am trying to get back to speed.

Soji
So, KEEP CALM and WAIT FOR NEW POLYGON CREST! ;-)
/!\ READ YSFLIGHT HANDBOOK (.PDF) | updated 2017/05/03

U.S. Senator John McCain wrote:
“Spending #NewYearsEve w/ brave #Ukrainian Marines at a forward combat outpost - we stand w/ them in their fight against #Putin's aggression”
http://pic.twitter.com/vkz0gdozVV
(31 DEC 2016)

User avatar
u2fly
Veteran
Veteran
Posts: 1191
Joined: Fri Oct 02, 2015 9:27 am
Favorite Aircraft: Antonov An-225 Mriya
Location: Ukraine
OS: Linux
Has thanked: 669 times
Been thanked: 267 times
Contact:

Re: New things from Soji

Post by u2fly » Sat Nov 26, 2016 6:30 pm

Soji Yamakawa wrote:Casting Shadow with the Shadow-Map Method
(November 26, 2016)

In YSFLIGHT for OpenGL 2.0, I have implemented the shadow-map method. I believe it can also be done with Direct X9, but, I don't want to spend even a second to learn an API that runs on only one operating system, and even declared deprecated by the vendor. Shadow-map feature will be available only in YSFLIGHT on OpenGL 2.0/ES 2.0 for now. If Vulkun API catches up the speed, I will add support for Vulkun. Then the shadow- map feature will be available in YSFLIGHT on Vulkun then. But, I don't know how popular Vulkun can become. It may just fade away. Right now, there is no replqcement of OpenGL 1.1. I need to go very high- level, which is pretty much useless for me, or very low-level, which needs work to make it useful for me. Not everyone needs photo-realistic movie- like graphics. For me, I need 3D graphics to prove my concept in my research. OpenGL 1.1 was just good for my purpose. To use OpenGL 2.0 with ease, I wrote a cover library. Maybe I'll need to do the same for Vulkan eventually.

But, what I desperately want to avoid is to spend considerable time to lean how to use Vulkun only to see Vulkun just fading away. I wasted too much time to learn Direct3D9. Microsoft ditched it in Direct X10. If I knew my knowledge in DirectX9 becomes useless in such a short time, I definitely wouldn't have wasted time learning Direct3D9. I learned from it. I will never spend time learning Direct X12.

The shadow-map method is a standard method for rendering shadow in the recent 3D graphics applications. However, most of the 3D- graphics techniques are not good for rendering a scene that expands up to 20 miles. First I need to render a texture in which each pixel is a distance from the light source. But, even if I use 2048x2048 texture, one pixel corresponds to 0.01 mile. A shadow of a F-18 becomes a square. Useless.

The solution is to use multiple shadow-maps. I have tried a method that is said to be the most efficient method to this day called Cascaded Shadow Map (CSM) with three shadow-map textures. But, it turned out still too coarse. Useless. If you have a high-end GPU, and if I can increase the shadow-map count to like 8 or 10, it may be useful. But, my goal is to keep YSFLIGHT comfortable with 3-year old hardware. Three shadow-maps seems to be the maximum, based on the performance on 2014 Mac mini. Actually, this method has a problem, too. Imagine a formation passes above you from the behind. The light source is also to the back far above. The shadow of the formation must be visible before the formation comes in to the view. However, if I implement the CSM method as described in the paper, the shadow won't be visible until the formation is in the view frustum.

In conclusion, I need to adjust the light volume depending on the view mode. CSM might be generally good solution, but if I am looking at a particular airplane, I want the maximum resolution around the airplane, not close to the view point. I haven't implement this adaptive light- volume yet, but I tested the implementation with a very ad-hoc light volumes.

By the way, I also reduced the shaking of the rendering when the view point is far from the origin. The problem was that the GPU calculates matrices in 32- bit floating points. I have modified my code so that CPU calculates the transformation matrices using 64-bit floating points. This change essentially gets rid of the shaking. A few years ago everyone was talking about General Purpose GPU. But, GPU vendors are stubbornly refusing to support 64-bit floating points. 32-bit floating point is too inprecise for serious applications. GPUs are pretty much useless for many engineering applications, until GPUs are capable of calculating 64-bit floating points.

Nonetheless, maximum Z-buffer precision used to be 16-bit. Now, it looks like the low-end GPU supports at least 24-bit Z-buffer. I am thinking to remove "Z-Buffer Quality" option in YSFLIGHT and simplify the rendering code.

By the way, I have tried machine learning to fly an airplane in YSFLIGHT, too. I tried to make a speed-controller with machine learning. Unless my machine-learning code can control the speed, there is no way to do more complex flying with it. The result was a failure. I let an airplane fly in many different conditions and used flight records as learning data. Then, I made my machine-learning code to control the airspeed. But, it didn't work well. I guess I need to study more about machine learning. Eventually I can make something. Let's see.
Image
Image
Image
/!\ READ YSFLIGHT HANDBOOK (.PDF) | updated 2017/05/03

U.S. Senator John McCain wrote:
“Spending #NewYearsEve w/ brave #Ukrainian Marines at a forward combat outpost - we stand w/ them in their fight against #Putin's aggression”
http://pic.twitter.com/vkz0gdozVV
(31 DEC 2016)

NightRaven
Senior Veteran
Senior Veteran
Posts: 1843
Joined: Tue Feb 07, 2012 10:00 am
Favorite Aircraft: .
OS: windoze 10 64-bit
Has thanked: 528 times
Been thanked: 606 times

Re: New things from Soji

Post by NightRaven » Sun Nov 27, 2016 5:30 am

Holy shit that is a real step up from the graphics on YS currently. Hopefully that'll be done soon :)
fk dis place kthxbai

User avatar
Vic Viper
Staff
Staff
Posts: 494
Joined: Thu Feb 24, 2011 7:01 pm
Favorite Aircraft: F-22A Raptor
Location: -9° 59' 6.90", -138° 49' 35.51
OS: Fail10PremiumEdition
Has thanked: 296 times
Been thanked: 180 times
Contact:

Re: New things from Soji

Post by Vic Viper » Sun Nov 27, 2016 5:42 am

What's with the hating on Vulkan? Open gl (at least the older ones he uses) is the one with support getting dropped soon. Thought he'd know better
Last edited by Vic Viper on Sun Nov 27, 2016 6:02 am, edited 3 times in total.
Image
VFA-49 FORUM

"If you don't know who I am...then maybe your best course would be to tread lightly."

User avatar
Nodoka Hanamura
Veteran
Veteran
Posts: 923
Joined: Thu Apr 07, 2011 8:01 pm
Favorite Aircraft: Su-24, Mig-29, Su-27, F-18, F-15, ADF-01, Icon A5
Has thanked: 513 times
Been thanked: 266 times

Re: New things from Soji

Post by Nodoka Hanamura » Sun Nov 27, 2016 5:45 am

By the way, I also reduced the shaking of the rendering when the view point is far from the origin. The problem was that the GPU calculates matrices in 32- bit floating points. I have modified my code so that CPU calculates the transformation matrices using 64-bit floating points. This change essentially gets rid of the shaking. A few years ago everyone was talking about General Purpose GPU. But, GPU vendors are stubbornly refusing to support 64-bit floating points. 32-bit floating point is too inprecise for serious applications. GPUs are pretty much useless for many engineering applications, until GPUs are capable of calculating 64-bit floating points.
And that is how letters get shit done. :lol:

But yeah, Good to see Soji applied that fix successfully. Hopefully we'll see a new build before the end of the year or by next spring.
█████████████████████████████████████████████████████████████████████
Image

User avatar
Flake
Core Staff
Core Staff
Posts: 4266
Joined: Thu Feb 10, 2011 8:47 pm
Favorite Aircraft: Boeing F/A-18A
Location: Australia
OS: Windows 10 (x64)
Has thanked: 814 times
Been thanked: 1159 times
Contact:

Re: New things from Soji

Post by Flake » Sun Nov 27, 2016 6:59 am

Casting Shadow with the Shadow-Map Method
Ooooo I think I know where this is going. Sun/Moon?
By the way, I also reduced the shaking of the rendering when the view point is far from the origin. The problem was that the GPU calculates matrices in 32-bit floating points.
Called it.

User avatar
u2fly
Veteran
Veteran
Posts: 1191
Joined: Fri Oct 02, 2015 9:27 am
Favorite Aircraft: Antonov An-225 Mriya
Location: Ukraine
OS: Linux
Has thanked: 669 times
Been thanked: 267 times
Contact:

Re: New things from Soji

Post by u2fly » Sun Nov 27, 2016 7:16 am

xX_SuperSic58_Xx wrote:
By the way, I also reduced the shaking of the rendering when the view point is far from the origin. The problem was that the GPU calculates matrices in 32-bit floating points.
Called it.
Why Soji still not implement for scenery geographical coordinate system (as "globe" with 0...360 deg longitude and -90...+90 deg for latitude ) that is continued instead of current center-oriented coordinate system (now its is "plane" coordinates are -∞...0...+∞)???
/!\ READ YSFLIGHT HANDBOOK (.PDF) | updated 2017/05/03

U.S. Senator John McCain wrote:
“Spending #NewYearsEve w/ brave #Ukrainian Marines at a forward combat outpost - we stand w/ them in their fight against #Putin's aggression”
http://pic.twitter.com/vkz0gdozVV
(31 DEC 2016)

User avatar
decaff_42
First Class Membership
First Class Membership
Posts: 3413
Joined: Sat Oct 29, 2011 7:23 pm
Favorite Aircraft: SR-71 Blackbird
Location: Eastern United States
OS: Mac OSX
Has thanked: 537 times
Been thanked: 1192 times
Contact:

Re: New things from Soji

Post by decaff_42 » Sun Nov 27, 2016 2:07 pm

It is probably because you will run into a similar problem when really far away. The radial coordinate would be something like 356.34543463453..... You would still eventually run out of floating point precision, just in a different way.
Check out my website YSDecaff for all my tutorials and addon releases.
Visit by Bitbucket page for Python codes: https://bitbucket.org/decaff42/

Feel free to ask me a question! I'll try to answer it as well as I can or recommend a more knowledgeable member if necessary.

User avatar
m88youngling
Contributor
Contributor
Posts: 117
Joined: Thu Nov 07, 2013 11:54 pm
Favorite Aircraft: R-9A Arrowhead - F-16C Rainbow Falcon
Has thanked: 38 times
Been thanked: 42 times
Contact:

Re: New things from Soji

Post by m88youngling » Sun Nov 27, 2016 4:35 pm

Sun/Moon in YSF would be pretty C:
Maybe even modifiable textures for them, and combined with OYS you could get moon phases over time.
Image
SARF Executive Officer (2IC)
[email protected]

User avatar
Neocon
First Class Membership
First Class Membership
Posts: 3748
Joined: Wed Oct 19, 2011 5:01 am
Favorite Aircraft: Baron 58
Location: Tennessee
OS: Win Vista
Has thanked: 1534 times
Been thanked: 1661 times
Contact:

Re: New things from Soji

Post by Neocon » Sun Nov 27, 2016 4:38 pm

These are truly exciting times.

User avatar
u2fly
Veteran
Veteran
Posts: 1191
Joined: Fri Oct 02, 2015 9:27 am
Favorite Aircraft: Antonov An-225 Mriya
Location: Ukraine
OS: Linux
Has thanked: 669 times
Been thanked: 267 times
Contact:

Re: New things from Soji

Post by u2fly » Sun Nov 27, 2016 6:34 pm

decaff_42 wrote:It is probably because you will run into a similar problem when really far away. The radial coordinate would be something like 356.34543463453..... You would still eventually run out of floating point precision, just in a different way.
No, in geographical coordinates there no meaning for "far away", because its a sphere and it hasn't start or end.
Only thing needed to fix there is precision (0.01 or 0.00001, etc.) of YSFlight position. And it's possible use fixed-point numbers instead of floating there.
/!\ READ YSFLIGHT HANDBOOK (.PDF) | updated 2017/05/03

U.S. Senator John McCain wrote:
“Spending #NewYearsEve w/ brave #Ukrainian Marines at a forward combat outpost - we stand w/ them in their fight against #Putin's aggression”
http://pic.twitter.com/vkz0gdozVV
(31 DEC 2016)

User avatar
decaff_42
First Class Membership
First Class Membership
Posts: 3413
Joined: Sat Oct 29, 2011 7:23 pm
Favorite Aircraft: SR-71 Blackbird
Location: Eastern United States
OS: Mac OSX
Has thanked: 537 times
Been thanked: 1192 times
Contact:

Re: New things from Soji

Post by decaff_42 » Sun Nov 27, 2016 6:49 pm

Okay I understand what you are saying now.

Basically I think it all comes down to what already exists. A change to a geographical coordinate system would require a massive overhaul of the code for maps, replays, and probably a lot more that I'm not thinking of. Without some sort of translation code, any old FLD would become useless overnight.
Check out my website YSDecaff for all my tutorials and addon releases.
Visit by Bitbucket page for Python codes: https://bitbucket.org/decaff42/

Feel free to ask me a question! I'll try to answer it as well as I can or recommend a more knowledgeable member if necessary.

User avatar
u2fly
Veteran
Veteran
Posts: 1191
Joined: Fri Oct 02, 2015 9:27 am
Favorite Aircraft: Antonov An-225 Mriya
Location: Ukraine
OS: Linux
Has thanked: 669 times
Been thanked: 267 times
Contact:

Re: New things from Soji

Post by u2fly » Sun Nov 27, 2016 7:26 pm

decaff_42 wrote:Okay I understand what you are saying now.

Basically I think it all comes down to what already exists. A change to a geographical coordinate system would require a massive overhaul of the code for maps, replays, and probably a lot more that I'm not thinking of. Without some sort of translation code, any old FLD would become useless overnight.
Yeah, I think this was major mistake of Soji when he firstly choose how scenery would be represented in YSFlight. But, as for me, now its a time to fix it as I described above. There no need change DNM or SRF coordinate system, but FLD and YFS ("YSFlight's world")
P.S.: "- Better (make this) late than never."
/!\ READ YSFLIGHT HANDBOOK (.PDF) | updated 2017/05/03

U.S. Senator John McCain wrote:
“Spending #NewYearsEve w/ brave #Ukrainian Marines at a forward combat outpost - we stand w/ them in their fight against #Putin's aggression”
http://pic.twitter.com/vkz0gdozVV
(31 DEC 2016)

User avatar
Neocon
First Class Membership
First Class Membership
Posts: 3748
Joined: Wed Oct 19, 2011 5:01 am
Favorite Aircraft: Baron 58
Location: Tennessee
OS: Win Vista
Has thanked: 1534 times
Been thanked: 1661 times
Contact:

Re: New things from Soji

Post by Neocon » Sun Nov 27, 2016 8:31 pm

A transition from a flat world to a globe would probably be worth the problems as long as it was possible to convert old maps to the new system in some way.

I just don't see it happening any time soon.

User avatar
Flake
Core Staff
Core Staff
Posts: 4266
Joined: Thu Feb 10, 2011 8:47 pm
Favorite Aircraft: Boeing F/A-18A
Location: Australia
OS: Windows 10 (x64)
Has thanked: 814 times
Been thanked: 1159 times
Contact:

Re: New things from Soji

Post by Flake » Sun Nov 27, 2016 9:07 pm

u2fly wrote:Why Soji still not implement for scenery geographical coordinate system
Because at the very least that would require another translation matrix and that would reduce render speed.

Honestly I think he's done for now. With a 64 bit decimal the maps can be much MUCH larger. We could make a world map finally.

Stormraider
Newbie
Newbie
Posts: 10
Joined: Sun May 08, 2016 12:56 am
Favorite Aircraft: B-1B Lancer
OS: Windows 10
Has thanked: 4 times
Been thanked: 4 times

Re: New things from Soji

Post by Stormraider » Sun Nov 27, 2016 10:06 pm

xX_SuperSic58_Xx wrote:
u2fly wrote:Why Soji still not implement for scenery geographical coordinate system
Because at the very least that would require another translation matrix and that would reduce render speed.

Honestly I think he's done for now. With a 64 bit decimal the maps can be much MUCH larger. We could make a world map finally.
Does that mean that Project Atlas will continue?

User avatar
Nodoka Hanamura
Veteran
Veteran
Posts: 923
Joined: Thu Apr 07, 2011 8:01 pm
Favorite Aircraft: Su-24, Mig-29, Su-27, F-18, F-15, ADF-01, Icon A5
Has thanked: 513 times
Been thanked: 266 times

Re: New things from Soji

Post by Nodoka Hanamura » Sun Nov 27, 2016 10:51 pm

Stormraider wrote:
xX_SuperSic58_Xx wrote:
u2fly wrote:Why Soji still not implement for scenery geographical coordinate system
Because at the very least that would require another translation matrix and that would reduce render speed.

Honestly I think he's done for now. With a 64 bit decimal the maps can be much MUCH larger. We could make a world map finally.
Does that mean that Project Atlas will continue?
What's Project Atlas?
█████████████████████████████████████████████████████████████████████
Image

Post Reply

Who is online

Users browsing this forum: Google [Bot] and 2 guests