Jeremy Miller tweets (Senior Programmer at ISI)

Here we go again... *Yawn*

 

Oh yeah I see... because the frequency of visits to the forum gives a completely reliable insight into how popular the sim is and how many are using it and what they think about it.

No wait. It doesnt. Sorry, my bad.

 

Forum visits don't, but online server populations do. It would be interesting to see a graph of those numbers and how they have changed over time. I only have a sense of those numbers based on my own experience... Based on my experience, it has been falling steadily. A decline in population would be expected after the 1.0 release of AC, since it is the new shiny, but the numbers appear have been falling since before then. Again, just my observations from the short time I have been involved in rF2. (about 8 months or so)

 

It's funny, I have noticed a significant increase in online server population this past month.
Tiger Moth and especially the NSX have got people racing again.
The Honda have really started a new increased interest for rF2 around the web it seems.
Really good stuff.

Before that things looked at bit grim tbh, but it's definitely picking up now.
People will grow bored with AC online pretty quickly I reckon, seeing how it's basically just rammers and kids dive bombing and wrecking each other.
Those races man... beyond travesty, I say...

 

RE: numbers in servers

Thus far we had one great season with the URD eGT mod, but it's died a death. So much so I have stopped the series early in order to try other things. I put this down to bigger "names" within the community running the mod, people putting their time in to those and thus ours is nigh on empty. This is fine by me, the community vote with their participation. It was suggested I mirror someone like VEC in terms of calendar but with sprint races, lessening the workload on people needing to do setups. I wasn't interested in going that route.

Our Clio series, while initially quiet, enjoyed a field 20+ strong last saturday evening. This is on again tonight and will continue to be while people remain interested.

Our BTCC series has done really well. Each meeting has in the upper teens of drivers and produces fun events.

I think, in part, it is up to the league admins like myself to recognize what others are doing within the community and try to offer something different and desireable. In part, not being afraid to try something new.

As part of this I operate an open server policy, they're never passworded. Some see this as a risk but I like the idea of anyone coming on the matchmaker on a friday or saturday night knowing that we will have something to race, should they choose. In an environment where they don't have to go and find a password and that everything downloads correctly from the server.

I don't see a decline at all but I don't really see an increase either. We're ticking a long nicely to a point where it is remaining viable to keep paying for a dedicated server, at least.

It's not as doom and gloom as a lot of people like to say it is.

 

Yes, we've had several open-server Saturday evenings recently & they've generally been good, & good fun too. But we do get maybe 2 reckless drivers out of 17-18, & they do have to be kicked out before the race starts..

So it does work if you're not afraid to operate "tough love"

 

For sure. I basically use my judgement. One act of.....odd behavior, I let it slide. On the second, they get a warning. On the third, no messing, IP ban.

Last week in the Clios I IP banned Four guys. Two were wreckers. The third and fourth I saw in chat "You're gunna get sent". I looked at the replay, they were ramming each other. Goodbye.

Then we continued and had a brilliant race.

 

An amazing number of servers on mm list but id guess 75% empty mostly. Our league keeps ticking along catering for Australasian racers and are trying a combined league with our Malaysian friends. As content increases and as ac drivers look to challenge themselves as sim racers numbers will increase. Luckily we have some pretty awesome offline racing with rf2 to bide time.

 

Lol... best thing I read today

 

Question for Jeremy, or others who could answer: Is there a particular problem with shadow rendering in 880, or is it just my imagination? It's the old "shimmering shadows" problem, with very poor AA, but only on shadows. The rest of the scene looks normal as per AA and AF settings.

 

I've got terribly jagged shadows as well, in cockpit mostly. So ugly. Tried all shadow and AA settings.

 

Do you recall it better before 880, or just the same? I am praying 880 is an anomaly!

 

+1

Inviato dal mio SM-T705 utilizzando Tapatalk

 

I've had various shadow problems on most builds, mostly drawing as you go and also sawtooth but I'm pretty sure I haven't seen them this bad in the cockpit before. The shadows are the biggest graphical problem from my point of view. All that flickering and sawtooth and watching them being drawn as you go looks so amateurish and it's been going on long enough now.

 

The sawtooth-edged shadows do ruin the immersion, especially when the other AA is OK.

 

Shadow edges have nothing to do with AA settings.

First thing you should do is to post a screenshot in the right forum section and provide description of your shadow settings.

Shadows will always have sawtooth edges - that's how shadows work in most games. Getting rid of it takes a lot of additional development time and GPU processing power. Unless you're experiencing something severe. I cannot tell from your descriptions, so I can only advise you post screenshots in the issue reporting forum.

 

Thanks for the info. All settings maxed. Why wouldn't shadows get AA like everything else? As soon as you add any movement (either the thing casting the shadow like a tree, or from a moving cockpit driving down a track), the sawtooth shadow edges turn in to a mass of pixelated shimmering that looks like a 1990's video game. I hope we can do better than that.

Also, the problem typically goes away when the objects/shadows are closer to the user POV. So how do the shadows that are close get AA, but not in the distance if shadows always have sawtooth edges?

How do certain shadows in the cockpit have sawtooth edges, but others do not and look terrific (iRacing has this issue a lot)? Sounds like a performance/quality trade-off like all else to me. There should be an OPTION for proper high quality shadows, optimized, so those who want to spend the resources can and those who do not can have their sawtooths. You can with higher resolution shadows:

http://feedback.arma3.com/view.php?id=1353

http://www.overclock.net/t/563420/mass-effect-bad-shadows-and-aa

http://www.gamefaqs.com/boards/615805-the-elder-scrolls-v-skyrim/60975074

And by the way, my question was related to 880 versus earlier builds. Is your assessment that shadow rendering has not changed in 880 or other more recent builds? It appears to me that higher res shadows are broken, but I will go test the settings again to see if they make any difference.

 

I agree with AA remarks above, but this particular problem is easy to explain: the less pixels you have to represent a complex shape, the more shimmering pixels you'll have. A tree shadow far from you is only few pixels high so choice have to be make from the computing (projection). The camera move sightly, the choice of displayed pixels change, that's impossible to avoid AFAIK. The more it grow, the more it have pixels to be represented, so the more the complex shape is displayed more accurately and gain "stability". This are not really something coming from AA (in this case), just due to resolution of our monitor, and "choice" made by formulas. Also, I guess they put some limits where AA is more computed when near than too far. This is why we see moving quality on shadow in front of you. I guess

 

I am still curious as to why Shadow Blur OFF, FAST, OPTIMAL and QUALITY all give identical performance to shadows cast by objects.

Only way to get rid of the shimmering ugliness is to bludgeon it with super-sampling. Links I put in above show that other titles have similar issues. In this case, no one is appealing to the console audience or hardware, so we should have an OPTION.

Hopefully it is just temporarily disconnected in 880.

 

Antialiasing helps with geometry edges. Shadows are not geometry, they're textures.

Textures are not antialiased, but they can have filtering. Unfortunately, this is not true for shadowmaps - they cannot be filtered and even if you would filter them, this would not make shadow edges look any different. This is because what you see on screen is not shadowmap texture, but result of shadow test made using that texture.
I will not go into details, but any filtering or antialiasing we have in your graphics cards just doesn't apply to shadows. Game developers need to use some advanced techniques if they want to soften their shadow edges. The better the result, the more complex the technique required or more performance impact it has.

I recently played Deus Ex HR. It got me thinking - why people keep comparing rF2 to 90's games if this 2011 blockbuster is in many aspects behind rf2? Shadows, for example, are of far lesser quality than what you can see in rF2.

I think it's because people get so used to technology, that they forget where we were 3-5 years ago. But truth be told - technology is not some magic - it's limited like everything else. If we don't get smooth shadow edges in every game, then we're not there yet and we need to wait a few more years. Right now some more advanced engines perform blurring of shadows, but let's not forget that these are made by companies making big bucks and investing that back into development time. They've been doing this for years and are ahead of smaller companies. You only need to look at this article to realize just how much research has gone into antialiasing techniques at some game studios. Seriously, I wouldn't be surprised if they had one person working for a couple of years almost exclusively on subject of antialiasing

Personally, I think shadows in rF2 are programmed pretty good and the only problem is that some tracks (even ISI ones) don't have them setup very well.
You can try Croft circuit by feels3 - you will see shadows have much higher quality there.

No.
Shadows closer to observer use higher resolution shadowmaps. It's the same in pretty much every game that uses shadowmaps in open spaces.

This is why I commented, that you should post some information. Sometimes people try new track and have problems that they blame on recently installed build (because they didn't have these problems before). This is why first reaction is always to ask about circumstances
Also, what I mentioned about being used to technology - some people report normal behavior as bugs, simply because what they get is below their expectations, so they consider it an issue

It's important to "prove" somehow, that your claim is reasonable, so that it can make it to development team, and not just to the forum. This is why I also mentioned posting in forum section dedicated to bug reports. Show comparison with the same content and the same setting between current and previous build.