As for SteveW:

I find your pontificating

And me your's. You have no technical input or any ideas of you own to offer. I provided info for my clients and it's been picked out by others and I found I had to defend it. It upset you, and you don't understand it - tough.

 

Steve, even though you can prove to us that we are all mistaken, my eyes tell me sgss helps in dx9 when combined with 8xs aa. And as you will have noticed, I am not alone in this..

 

No disrespect, but I would suggest you stay with dx10 where you are happier, and we die hard dx9 fans continue to muddle through on our own.

I agree.. 8xs and 2xsgss in dx9.

Does not tally with:

 

 

Ok, set that to 8xS please Steve. You need to feed SGSS 4x multisamples for it to work (theoretically you need 2x multisamples for 2xSGSS, 4xMSAA for 4xSGSS and so on). It's eather plain 4x MSAA or some combined mode that includes 4xMSAA like 8xS

Maybe should have left it Bert, sorry if this causes any embarrassment, but your colleagues do not agree! You're all at sea.

 

It's quite clear from this discussion folks are making more than one change at a time and misleading themselves, e.g. "8xs and 2xsgss".

 

 

To the OP. Sorry this became some kind of debate about my recommendations - that was not my intention. On the same day I was solving other's problems in quick time. I'm only here to help you.

Does not tally with:

 

 

 

Maybe should have left it Bert, sorry if this causes any embarrassment, but your colleagues do not agree! You're all at sea.

 

It's quite clear from this discussion folks are making more than one change at a time and misleading themselves, e.g. "8xs and 2xsgss".

 

 

To the OP. Sorry this became some kind of debate about my recommendations - that was not my intention. On the same day I was solving other's problems in quick time. I'm only here to help you.

I told you that you needed 4 multisamples, theoretically, to run 4xSGSS. 8xS is supposed to include 4xMSAA so it should run both 4 or 2x SGSS. Not sure what you mean by all that. The only one who is alone in this thread is you. 8xS + 2xSGSS works fine as well. 

Well seems obvious that Bert's recommendation does not agree with yours. You're all at sea.

That's fine that the thread took a different turn. While its off topic and there's dx10 discussion, can anyone point me in the right direction on how to convert dx9 to dx10 as far as settings for inspector, fsx, Cfg file. Any links appreciated.

arnobg try this 

 

http://#####.wordpress.com/fsx-software-and-hardware-guide/

 

Hope you get it sorted

point me in the right direction on how to convert dx9 to dx10 as far as settings for inspector

The same link I posted above (causing unnecessary consternation for a few) demonstrates the requirements for NI with DX10. Primarily the "Antialiasing - Transparency Supersampling" dropdown selection is used instead of the "Antialiasing - Setting" dropdown because DX10 is different to DX9. But please see my link above. Also you should at least install the shader_release_v3.2.3 (see Misc section AVSIM File Library) which fixes some issues with DX10 rendering. However some scenery and planes may not work fully with DX10 but we can get the SteveFX scenery fixer program, this cures most DX10 issues with addons.

I told you that you needed 4 multisamples, theoretically, to run 4xSGSS. 8xS is supposed to include 4xMSAA so it should run both 4 or 2x SGSS. Not sure what you mean by all that. The only one who is alone in this thread is you. 8xS + 2xSGSS works fine as well.

As I mentioned previously, the Supersample settings don't change the view with SG because they are already supersample. Adding a set of flags to produce SG in Transparency mode while using FSX DX9 just happens to set Supersampling where there are multisampling values in the dropdown.

 

But anyway I tried all modes. With combined modes e.g. "16xS [Combined: 2x2 SS + 4x MS]" with "2x Sparse Grid Supersampling" took pictures, zoomed in, and in every case there is a very small, but odd, loss of detail. Like some pixels are never visited in the sample, and a slightly blurred result overall. In each case the combined modes were better left alone. With SG, the Supersample only modes showed no difference, the Multisample only modes mimicked the Supersample modes, and the combined AA modes all looked worse. For such a small difference, obvious lack of detail, and if it's sapping the fps so bad, why use it?

 

BN2 VC, combined mode on the left, added SG on the right:

 

 

 

Whereas SG used in FSX DX10 and P3D DX11 make a massive difference when needed.

 

SSAA on the left appears non existent with the BN2 VC FSX DX10, SGSSAA on the right:

 

 

EDIT: In the shots, while in the sim, I zoomed fully out and moved the eye-point back until the numbers on the instruments were just visible and took the shot (V) in 1920x1200. I plucked a small area from each and zoomed in on those.

Bingo, you see it does work.. and now the debate is about why bother use it.. where it should have been all along.

 

I use it because it reduces shimmering in the cockpit.

 

And if that means losing some detail, it is a worthwhile tradeoff for me.

 

This does not mean that everyone should use it.. it is a choice.

Bingo, you see it does work.. and now the debate is about why bother use it.. where it should have been all along.

Bert you deliberately miss the point - sad. Several times you only admit to using 8xS + SG which, as has been mentioned, several times, there must be an "M" for Multisamples in the mixture anyway for the "effect" to show.

You understand correctly... I use 8xs as the baseline, and optionally add 2xsgss.

Yep. 8xS + SGSS looks better (the more SGSS the better it looks) than any combined (xS) mode because for some reason, the combined modes' supersampling doesn't seem to apply to the full scene like SGSS clearly does. This is particularly apparent in autogen trees, so I suggest Steve, you try plain 8xS (or any other mode for that matter) and compare it to 4x or 8xSGSS in a close up on some AG trees

You understand correctly... I use 8xs as the baseline, and optionally add 2xsgss.

On my 1920x1200 panel, I had to zoom right out and move the eye-point backwards from the VC until the numbers on the small instrument could just be made out. Then as stated, with the combined SS + MS modes, the most tiny differences could be made out, as shown in the smaller image above. I have to look closely for any changes at all, and then, there are irregularities in the highlight on the curved instrument edge that were not evident with the SG off. There were no real differences with shimmering at this level of zoom with either case anyway, and the application of SG seemed pointless.

 

Whereas, experimentation with the DX10 and DX11 modes prove very worthwhile changes with SG, but only on some models. The stock Baron, for example, looks no better with SG.

 

I kept the information about Multisamples back for a day but no-one came up with it. However there are plenty of posts around the forums that are attuned to the proper use of SG in the transparency section of NI with some aircraft in DX10 and DX11. The effect in DX9 is just a side effect unfortunately, and not the result of a sparse grid sampling mode.

 

Anyway I can say whichever camp you're in doesn't really matter. I've brought a lot to the table concerning these settings, very little from anyone else. I'm sure you and other picked up a few pointers. All the best Bert, and happy flying - that's the main thing.

compare it to 4x or 8xSGSS in a close up on some AG trees

Yes I see the same tiny difference can be made out in the trees, but seems not to impact with improvements, just a tiny difference. The trees at those zooms seem much harder to compare than the edges of ultra-zoomed in instruments, and it's not cleared anything up in that respect. I set the aircraft up in a saved flight so that the same starting point can be repeated and compared accurately.

The trees at those zooms seem much harder to compare than the edges of ultra-zoomed in instruments

 

If you have problems zooming in, just look out the window at a normal zoom level and in motion. The shimmer in trees is so obviously killed almost completely at 4xSGSS, let alone 8xSGSS that it's impossible to ignore.