FS2004 Increase number of autogen?

[Navid Hashemi]

Hello developers!

Is there any way to increase the number of autogen? For example if i want the forests more dense.


greetings
Navid

 

[GaryGB]

Hello:

Edit a [FS2004 install path]\FS9.Cfg parameter under the [TERRAIN] sub-section:

[TERRAIN]
TERRAIN_AUTOGEN_DENSITY=5 <-- values up to 30 are known to work OK in FS9 for displaying more trees

[EDITED by GaryGB]
NOTE: this should read:

[TERRAIN]
TERRAIN_AUTOGEN_DENSITY=5 <-- values up to 30 has been reported to work OK in FS9 for displaying more trees

[END_EDIT]



NOTE: One must do this edit (ex: via a batch file system) to a higher value before starting FS2004 each time, as FS9 resets back to default of "5" when it exits.

BTW: TERRAIN_AUTOGEN_DENSITY value > "5" visually maxes out the autogen density slider (don't adjust autogen slider after start FS9 with value > "5" in FS9.Cfg)


Alternatively, in a "custom" land class scenery project, one can use the FS SDK "Autogen Annotator" utility to create and edit the types of autogen and the density at which it displays (up to the limits of the FS2004 display engine... which IIRC, is approximately 1/10th the density achievable in FSX).



FYI: A different set of variables and rules apply in FSX, so this is for FS9 only !


Hope this helps... and Happy Flying !

GaryGB

 

[rhumbaflappy]

I think there was a texture set that replaced the default autogen trees with clumps of trees. That would increase the number of individual trees... but not necessarily the density, as the autogen placement points would remain the same.

Dick

 

[luap]

Hello:

Edit a [FS2004 install path]\FS9.Cfg parameter under the [TERRAIN] sub-section:

[TERRAIN]
TERRAIN_AUTOGEN_DENSITY=5 <-- values up to 30 are known to work OK in FS9 for displaying more trees
GaryGB

Is this true with default textures? I thought agn was a function of how the textures were annotated. Will more agn polys need to be added or it works straight away?

 

[ALain152]

Hello:

Edit a [FS2004 install path]\FS9.Cfg parameter under the [TERRAIN] sub-section:

[TERRAIN]
TERRAIN_AUTOGEN_DENSITY=5 <-- values up to 30 are known to work OK in FS9 for displaying more trees

NOTE: One must do this edit (ex: via a batch file system) to a higher value before starting FS2004 each time, as FS9 resets back to default of "5" when it exits.


GaryGB

Hi,
comming back to that thread I tried to populate some LODs with 300,600,1200, and 2400 trees.
I couldn't find any value of TERRAIN_AUTOGEN_DENSITY up to 30 that changes anything in the number of trees displayed". FS9 displays the first 600 trees then leaves the rest blank

So 5 seems to be the actual max value for TERRAIN_AUTOGEN_DENSITY and 600 trees the max value / LOD

 

[GaryGB]

Hello:

Edit a [FS2004 install path]\FS9.Cfg parameter under the [TERRAIN] sub-section:

[TERRAIN]
TERRAIN_AUTOGEN_DENSITY=5 <-- values up to 30 are known to work OK in FS9 for displaying more trees

Hello:

I posted in haste that day (post-haste ? ); I have edited my post above.

I should have written "has been reported" rather than "known"; sorry if I gave the wrong impression.


I too had read FS2004 SDK info stating the limits on autogen objects per LOD13 quad (alternatively referred to as a "tile", "area" or "cell").

IIUC, in FS9, the LOD13 quad is commonly regarded as the smallest terrain tile quad matrix unit one can work with for custom 3rd party scenery, understanding too that there are 256 x 256 "area points" resolvable as Lat/Lon coordinates by the FS9 rendering engine within a LOD13 "area".

There is of course an apparent FS9 "Limitation on Autogen Trees":

http://www.newsite.fsdeveloper.com/forum/showthread.php?p=95608


I would like to offer some related info here as a suggestion of something to be further evaluated by any interested party willing to take the time to do properly detailed and/or quantitative testing.


I first encountered this idea of possibly increasing FS9 autogen here in 2007:

"Make Your FS2004 Autogen Dense Like FSX!"

http://forums.simflight.com/viewtopic.php?f=48&t=62445

Sometime in the subsequent 2 years I began experimenting with tweaking various FS9.Cfg settings, and noted a subjective impression (non-quantitative) that with TERRAIN_AUTOGEN_DENSITY settings greater than 5, there "appeared" to be an increase of displayed autogen in FS9.

However, at the same time I was also tinkering with "experimental" (and generally very large) numeric FS9.Cfg parameter settings that greatly increased the visual display radius for terrain mesh, as well as "sharpness" of vector content, and the perceived speed at which terrain mesh and autogen objects were resolved and rendered on-screen by the FS9 rendering engine.

Those "tests" were performed on my older FS development and testing computer that used a AMD Athlon 3000+ single core CPU on a 5% overclocked FSB, with 1 GB RAM set at 333 MHz / minimal CAS latency, and a 15% overclocked ATI A.I.W. X800 XT AGP-8x video card with 256 MB VRAM / 256 MB AGP Aperture running at 1280x960x32-bit; that system runs with a bare minimum of Windows XP Pro services and processes, having PCI latency individually adjusted to the minimum for all PCI cards (and in particular, latency="0" for that AGP video card and driver version).

FYI: That older FS computer easily achieves 48 FPS in FS9 (no AI Traffic, no Weather/Clouds="Simple", no Glare/Flare, no FS Anti-Aliasing /AA via drivers) with all other sliders/settings maxed.

In FSX that "legacy" computer achieves 24 FPS (no AI Traffic, no Weather/Clouds="Simple", no Glare/Flare/Bloom, no FS Anti-Aliasing /AA via drivers), with all other sliders/settings maxed except LOD Radius="Medium", Water="Low 2.x", Autogen Density="Dense".


It is possible that the other tweaks I had applied in FS9.Cfg (by increasing the visual display radius and speed of scenery rendering) may have minimized the latency for the alpha-fade-in for autogen objects, and allowed a subjective perception that one was seeing "more" autogen local to the aircraft.

Perhaps this was due to the increased visual display radius allowing display of more autogen objects in LOD 13 tiles outside the immediate tile that the aircraft was over at a given point in time.

At that time, I personally had not yet gotten around to testing a setting higher than TERRAIN_AUTOGEN_DENSITY=14.

However, I shared this info with some FS9 enthusiasts who maintained "stock" installations of FS9, and flew almost exclusively in Holger's Glacier Bay area, as they were wanting to test some tweaks that might enhance their FS9 experience even more than they were already enjoying on computers of various ages and levels of performance.

I mentioned the curiosity of apparent increased display of autogen when certain settings were used in a "test" copy of FS9.Cfg substituted for their current FS9.Cfg, and said I'd be interested in hearing whether in their "tests" they saw any further apparent changes at higher settings of the TERRAIN_AUTOGEN_DENSITY parameter.

They consistently reported similar findings of a "perceived" increase of autogen trees in the areas they were flying, and I understood them to say that they noted perceptible changes in displayed autogen trees up to a setting of TERRAIN_AUTOGEN_DENSITY=30.

IIRC, they were also using the other FS9 "experimental" tweaks I was testing that greatly increased the FS scenery visual display radius around the aircraft.

So I suppose one might wonder if those FS9.Cfg multi-parametric changes enabled the subjective impressions of "increased" autogen that they reported.


Thus the history behind the "hearsay" (heresy ? ) I forwarded in good faith, "post-haste", to offer an idea to the OP in this thread.



As Dick has pointed out above, there have been a number of contributions to the FS2004 Community which endeavored to change the autogen tree textures displayed on a single tree autogen footprint to create a "stand", "copse", group" etc., thereby increasing the apparent number of trees.


[EDITED]
There were also some contributions made which increase the "apparent" number of trees displayed in FS9 by substituting custom annotated versions of FS system default-named autogen *an[x].agn files.
[END_EDIT]

Others re-sized tree textures to smaller, "more life-size" versions for certain tree types (although sizes of some tree types reportedly were actually true to life according to relatively recent statements by certain FS-insiders).

Still others re-arranged the placements of autogen footprints, with or without changing footprint density or textures.

Most of these files are available as freeware downloads via AVSIM, Simviation, Flightsim.com etc.

If I get some time, I'll edit this post to add a list of such autogen tree packages, and link to their download URLs.


One (freeware) autogen add-on which I previously enjoyed using (only available via the author's website) is:

AUTOGENIUS 2004

http://www.aliditalia.com/autogeniusengl.html


A recent (payware) add-on which works in both FS9 and FSX is:

SAMOSHIN - NATURAL WORLD TREES V2

http://secure.simmarket.com/samoshin-natural-world-trees-v2.phtml

IMHO author Alexei Samoshin more closely represents the appearance and density of conifer and birch tree types I see in areas where I do most of my flying.


But certainly there are a number of other excellent freeware and payware FS9 add-ons that are noteworthy I have not mentioned here (as of yet).



I hope this info helps the OP, and might serve as stimulus for discussion on how one can further maximize the experience of FS9... as an excellent "alternative FS" still to be enjoyed when one is not flying in FSX.



PS: IIRC, even if one annotates an FS2004 *.agn file with more than the "limits" in trees or houses, the "surplus" foot prints are ignored by the FS9 rendering engine, and IIUC such "over-annotation" therefore causes no problem to the FS9 rendering process or performance.


[EDITED]
But one's labors placing autogen "beyond the FS9 limits" need not be regarded as a waste of time, since IIRC, FS2004 format *.agn files appear to be rendered by FSX additively when placed into the [FSX install path]\Scenery\World\Texture folders along with *.XML versions of the FSX autogen files having *an[x].agn file names ("x" being the numeral 1,2,3,4,5,6, or 7).

NOTE: These FSX autogen *an[x].agn file names are reportedly the same file names for the FS world default "non-custom" land class names as are used in FS2004).
[END_EDIT]


NOTE: IIRC, to test this, one would need to create an entry in FSX Scenery.Cfg that adds the [FS2004 install path]\Scenery\World\Scenery folder as an "active" area, so the FSX rendering engine can access the FS9 autogen tree textures in the locally paired [FS2004 install path]\Scenery\World\Texture folder at run time.


FYI: I haven't yet tested "quantitatively" in FSX using FS2004 *.agn files containing annotations beyond a declared FS9 rendering engine display limit.

Naturally, I might be interested in hearing whether, in anyone's "quantitative tests" using a "double-autogen file" scenario, they saw any further 'increase' in displayed FSX autogen tree density... independent of the FSX.Cfg parameters for TERRAIN_MAX_AUTOGEN_TREES_PER_CELL !

....Or NOT !!!

http://forums1.avsim.net/topic/279576-qestion-re-holgers-autogen-tweak/



PS: Since AVSIM deletes images from threads such as the one linked above, usually within a weeks time, this often results in a loss of information vital to the further interests and knowledge base of the FS Community, so I have posted an archive of that thread below.

This archive is posted for download in a ZIP file; once un-ZIPped, to open in one's default browser, double-click the *.HTM file name in Windows Explorer.

The archive apparently has all content, including images and links from the original web page intact and unaltered, as created on 05-01-2010 via a default function in FireFox by clicking:

File > Save Page As > [thread file name.htm] > Save As Type: Web Page, Complete

I doubt there is a legitimate basis for anyone objecting to my posting this in good faith to be downloaded here, as I do so only for the FS Community to better understand graphically the interaction of autogen annotation density vs. FS Cfg autogen parameter tweaks vs. Autogen Density slider settings; but I defer in advance to any moderator discretionary removal.

I'd instead offer it to anyone wishing to receive a copy here via PM, but thus far, FSDeveloper doesn't seem to allow sending PM's with attachments.


[EDITED]
BTW: The "official" MS KB article on the Autogen attenuation tweak is here:

http://support.microsoft.com/kb/555739

[END_EDIT]


I'm curious if some / all of these parameters might be applicable to FS9 as well as FSX, which is of course, a basis for the thread above.

http://www.fsdeveloper.com/forum/attachment.php?attachmentid=8147&stc=1&d=1284049512

GaryGB

 

[GaryGB]

Hi All:

I just stumbled across this thread which was subsequent to a PM'd reply I sent in response to to an inquiry by someone:

http://nzff.org/forum/lofiversion/index.php/t11840.html


I thought some might find it interesting...

GaryGB

 

[GHD]

Does no-one read the SDK?

From the FS9 SDK

Limitations in Flight Simulator

For performance reasons, the scenery engine in Flight Simulator limits the number of buildings displayed to 300 per square kilometer and the pieces of vegetation displayed to 600 per square kilometer.
Future versions of Flight Simulator, however, may increase these limits; there is no limitation on the number of building footprints and vegetation areas stored in the annotation files.

Or perhaps they refuse to believe it

 

[GaryGB]

Hi George:

IMHO, it would be even more interesting to know what portion of the "building footprints and vegetation areas stored in the annotation files" end up getting rendered (...and why), when an FS annotation file's potential density of "requested" autogen exceeds the limits of a given version of FS for the number of buildings and vegetation that is "allowed" to be displayed per square kilometer.


IOW, how is it decided at run-time as to "which" annotations of the surplus "requested" in ones file ...will actually be rendered ?

Thus, one might wonder how to best plan the "placement" of one's annotations to ensure the best result for a particular version of FS, knowing that some "surplus" data may ultimately not be rendered.

[EDITED]
And can one in FS9 also use BOTH the *.XML file and the annotation *an[x].agn file to increase "perceived" or actual autogen displayed at run-time (...just as one can in FSX) if both files are in the appropriate active FS add-on or default FS system sub-folder ?

[END_EDIT]

"If the world should blow itself up, the last audible voice would be that of an expert saying it can't be done."

GaryGB

 

[ALain152]

Hi Gary

...what portion of the "building footprints and vegetation areas stored in the annotation files" end up getting rendered (...and why), when an FS annotation file's potential density of "requested" autogen exceeds the limits of a given version of FS for the number of buildings and vegetation that is "allowed" to be displayed per square kilometer.

As far as I could test it, the portion displayed (when slider is set full right) is 100% of the first 600 trees and 300 houses written in the agn file.

 

[GaryGB]

Hi Alain:

Did you mean that FS processes autogen coordinate locations starting at the top of the annotation file, and renders only those first ["x"] entries (up to the ["x"] max per type ...buildings versus trees), then it disregards the remainder through the bottom of the file ?


Thanks for sharing your insights on this !


GaryGB

 

[ALain152]

Yes if the slider is 100%. Otherwise fs9 seems to display a part of the entire file. Making some trees invisible at 100% becoming visible.

 

[GaryGB]

Hi Alain:

Thanks for the reply... I must say I find this somewhat intriguing.


To explore a hypothetical scenario, suppose one creates a new annotation file with the FSX X-Pack SDK for a photoreal area that previously had no associated annotation file.

Then, suppose one manually applied autogen footprint coordinates for trees beginning at the East side of the imagery area, and progressed Westward in an even, linear pattern... right to left.


Let us also suppose that by the time ones applied annotations reach the West edge of the imagery area, one had placed twice as many autogen coordinates as FSX was "allowed" to render by ACES coding, assuming FSX GUI Autogen Density slider settings were set at 100%, and there was no TERRAIN_MAX_AUTOGEN_TREES_PER_CELL=["x"] "attenuation" tweaks applied in the FSX.Cfg.


Would FSX only display autogen trees 1/2 of the way across the rendered imagery from right to left, beginning at the East edge ?


This assumes annotations are stored top to bottom within the *.XML file and / or *.BGL file... in the order they were applied.


Or is there a different order annotation coordinates are sequenced into after Annotator saves / closes /compiles that *.XML "work" file ?


One might wonder whether this info might prove to be a consideration in one's work-flow as one annotates either manually, or with a utility via a semi-automatic or a fully automatic method... and there may end up being a substantial number of "surplus requests" for trees at annotation coordinates that FS will NOT allow to be rendered.


I'd welcome further input here as to what else one might need to know about this to better pre-plan one's annotation work-flow.

GaryGB

 

[ALain152]

Hi,
first of all I'm speaking about FS9. I didn't test with FSX.

Would FSX only display autogen trees 1/2 of the way across the rendered imagery from right to left, beginning at the East edge ?

No. It will display the first 600 found in agn which means the first written in the agn (annotated when designing)

This assumes annotations are stored top to bottom within the *.XML file and / or *.BGL file... in the order they were applied.

No. In FS9 they are stored in the order you design. For polygons I don't use and didn't test.

 

[GaryGB]

Let us also suppose that by the time ones applied annotations reach the West edge of the imagery area, one had placed twice as many autogen coordinates as FSX was "allowed" to render by ACES' coding, assuming FSX GUI Autogen Density slider settings were set at 100%, and there was no TERRAIN_MAX_AUTOGEN_TREES_PER_CELL=["x"] "attenuation" tweaks applied in the FSX.Cfg.


Would FSX only display autogen trees 1/2 of the way across the rendered imagery from right to left, beginning at the East edge ?

Hi,
first of all I'm speaking about FS9. I didn't test with FSX.

No. It will display the first 600 found in agn which means the first written in the agn (annotated when designing)

No. In FS9 they are stored in the order you design. For polygons I don't use and didn't test.

Hi Alain:

Thanks for the further reply.


IIUC, in this hypothetical test scenario, if one placed 1,200 tree annotations in a East to West direction and sequence into an FS9 area in which ACES' coding only allows 600 trees to be rendered at run-time:


1.) One might infer from your description of how FS9 SDK Annotator stores end-user autogen annotation tree coordinates:

a.) Autogen annotation tree coordinates are stored in the order that they are added to the Annotator intermediate work *.XML file (...and IIUC, also in the *.agn output file)

b.) FS9 will display ONLY the first 600 trees found in the *.agn output file (meaning the first 600 tree annotations written into the *.agn in the same order that they were added.)

2.) FS9 will display ONLY the first 600 tree annotations starting at the East edge, and will therefore "max out" 1/2 way across the FS9 area, since there are 1,200 total tree coordinates in the autogen annotation file for the area in which ACES only allows 600 tree to be displayed.


Thus, we might only see half of the trees we placed in our test autogen annotation file, and if the process works as you seem to describe, we'd see trees only on the East half of that FS9 imagery area.


So, my understanding would be, based on this info, we may wish to carefully consider the way we annotate for trees manually, and via semi-automated or fully automated methods using a annotation utility.


If we add trees in a linear direction from one side of the imagery to the other, rather than evenly throughout the imagery, when the "max" 600 tree limit of the FS rendering engine is reached in an autogen annotated area, we might see a "gap" in our autogen coverage.



Assuming this is, in fact, how this aspect of the FS Annotation process works, one might wonder whether or not we'd do well to apply our annotations evenly throughout the imagery area.


I'd certainly welcome additional input which might shed more light on this aspect of working with the FS Autogen Annotator SDK.

GaryGB

 

[ALain152]

Thus, we might only see half of the trees we placed in our test autogen annotation file, and if the process works as you seem to describe, we'd see trees only on the East half of that FS9 imagery area.


So, my understanding would be, based on this info, we may wish to carefully consider the way we annotate for trees manually, and via semi-automated or fully automated methods using a annotation utility.


If we add trees in a linear direction from one side of the imagery to the other, rather than evenly throughout the imagery, when the "max" 600 tree limit of the FS rendering engine is reached in an autogen annotated area, we might see a "gap" in our autogen coverage.

That's it
It's fairly easy to make your own tests

 

[GaryGB]

Thanks again for all your feedback, Alain !

PS: I just stumbled across another (older) thread raising questions about "perceived" increases in displayable autogen trees in FS9.x:

http://www.visualflight.co.uk/forums/topic.asp?TOPIC_ID=2212

"FS2004 SDK documentation concerning the limits may have been carried over from the FS2002 doc without anyone spotting that the limits had been raised? (Maybe)"

"I am sure the 600 tree limit must have changed then as, after checking my autogen areas again, I find most have over 600 trees per tile and many over 1000 with a few over 2000. - all displaying OK."

GaryGB

 

[GaryGB]

UPDATED February 13, 2013

To clarify the issue of FS2004 limitations in display of Autogen, please refer to the FS2004 Autogen SDK documentation, Page 51 of 57

"Limitations in Flight Simulator

For performance reasons, the scenery engine in Flight Simulator limits the number of buildings displayed to 300 per square kilometer and the pieces of vegetation displayed to 600 per square kilometer.

Future versions of Flight Simulator, however, may increase these limits; there is no limitation on the number of building footprints and vegetation areas stored in the annotation files."


Also, the term "Cell" is used in multiple contexts in the FS SDK with different meanings, especially within the FS Autogen SDK.


To clarify the definition of a "Cell" with regard to ex: FSX Autogen attenuation tweaks in the FSX.Cfg file to limit display of autogen to a level less than the default rebdered at a particular slider setting in the FSX GUI, please note that the "square Kilometer" referred to above is actually citing the Quad Matrix Grid size of a LOD-13 / QMID-15 quad.


FYI: In this context, the size of a "Cell" based on SDK calculations is a LOD-13 / QMID-15 quad, and is not a 'square Kilometer', but rather 1,223 Meters x 1,223 Meters.


"Area Points" within a LOD-13 / QMID-15 quad are 4.777343750 Meters x 4.777343750 Meters in size (1/256 of the span of a LOD-13 / QMID-15 quad).


BTW: In FS2004, no more than 1 pixel of a photo-real texture tile can be displayed over a LOD-13 / QMID-15 quad Area Point span on the ground as a terrain texture, thus we see the stated maximum resolution of FS2004 photo-real generated via FS2004 SDK Resample being 4.777343750 (aka "4.8") Meters per pixel.

GaryGB