Unity 3.51: still in beta with these bugs

[rextr09]

Say you have 1 cube. For each light that affects that cube, the cube is rendered again, hence the increase in draw calls, vertices and tris. I.e. 1 cube and 4 lights = 5 cubes.

Are you sure? I think it's still 1 cube but with 5 draw calls.

[n0mad]

Are you sure? I think it's still 1 cube but with 5 draw calls.

verts and tris are effectively multiplied by the number of lights hitting the object, 100% sure.

[_rem]

Excuse me Ippokratis, I'm a little exasperated.
I spent several days trying to solve these problems with the 3.5 and I thought that version 3.51, will solve all problems.
I use the Projectors, to create footprints and the shadow of the character, instead of generating 4 drawcalls by Projector (terrain with 4 textures) 3.50 and 3.51 generate over 60 drawcalls per Projector !
If I deactivated all the projectors drawcalls are identical with the version 3.42, but there are still performance problems in the Editor and webplayer.
I work with PCs rather old, with versions 3.5x, they are on their knees. XP says: Unity uses 150 mo but in reality it consumes between 600 and 700 mo. Unity 3.42 consumes only 150 mo.
I do not understand anything, I think if the guys of unity not find me a solution, Unity is over for me, I will not upgrade to Unity 4 !

[Ippokratis]

Hi,
In my opinion discussing about the problem will help you to sort it out.
Projectors are little beasts on their own, I never really liked them ( or shadows, or photorealism ) hence I never found my self having troubles with them :P

As Aras once said :

Projectors work by rendering all objects they "touch" with the Projector's material. What the projector's material does is completely open; the Projector just also provides some matrices so that the material+shader could project some texture, for example. But it could just as well just render object in pure red. So yeah, the major difference from "decals" is that projector renders full object, even if it touches a very small part. Pros of this approach: the only cost on the CPU is the draw calls. Cons of this approach: much more filrate consumed, especially if model is very large.

Joachim recently stated :

The framerate issues have been tracked down to 3 sources:

1) A bug in the Projector, causing a big increase in draw calls. This is fixed in 3.5.1

2) An intended change in behaviour of how lightmapped static objects or affected by realtime lights. This fixes a bug but some "incorrect" lighting rigs will get increased draw calls. This can be fixed easily in the project. Details are here:
http://forum.unity3d.com/threads/116...l=1#post833498

3) A performance regression in the skinned mesh renderer on some iOS and Android devices. This is due to some driver issues with multithreaded skinning. This is fixed in 3.5.1.

4) On Android specifically Adreno chipsets have very slow dynamic VBO performance. We added a specific driver workaround.

If you have a performance regression and it doesn't apply to those 3 cases, please file a bug and post the case number here.

We are in the final stages of testing 3.5.1. It will ship soon.

Unity 3.5.1 Fixes section : Graphics: Fixed projector culling so it only renders objects inside projector.

Alternatives :
This guy offers an alternative for shadows, have not tried it though.
Decals are also available ( many people have something bad to say about this package, I tried the free version and I am amazed on how clever these guys are and how good it works once I RTFM ).

[Joachim_Ante]

Do you have a case number for the bug you submitted on the performance regression, that includes a sample project where I can reproduce the issue?

I'd like to take a look at it.

[_rem]

Do you have a case number for the bug you submitted on the performance regression, that includes a sample project where I can reproduce the issue?

I'd like to take a look at it.

Thank you for your interest in my problems, you will find explanations and links on webplayers.
IF you think you need the my project folder, ask me here!

Case 458215

[antenna tree]

IF you think you need the my project folder, ask me here!

Submitting a sample project that reproduces the issue is always what we prefer to have so please add it to your bug report.

[BigBulle]

Yes but it takes time and it is really boring to do in a production process...

[Ippokratis]

Hi,
Another possible alternative is to give access to svn ( after prior arrangement with Unity ), for huge projects.
Thing is it is not possible to spot errors without the source. It takes time, is really boring but eventually helps you and rest of Unity community.
-Ippokratis

[_rem]

Hello,
I decided to create a new project and re-import all my "assets" one by one. After several hours fighting with broken links, "levels", "Tags", hardlinks between scripts and "gameobject", scripts and scripts, remake "ligthmaps" ... for short:
- Everything works perfectly, no "dracalls" crazy, no display problem with "the prefab Nighttime Water", great! the same performance as version 3.42.
I decided to apply "Occlusion Culling" on my "terrain" and on elements of decor, I check with "visualization" everything is OK, I "Bake".
And there, all problems begin! "dracalls" crazy, refraction of "prefab Nighttime Water" blinks, the framerate collapses ...
I delete data "Occlusion Culling" and "Occlusion Area" and everything returns to normal, I create a "webplayer" the same performance as version 3.42.
- "Occlusion Culling" not work with 3.51 (it did not work, also with 3.42 but silently)
I'll edit my bug report with a simplified project, (I found another strange big bug not easy to reproduce!)
thank you for the support.

[l0cke]

Unity 3.5.1:
-When turning on text asset serialization it works for 1/2 hour (makes sense, big project) and then out of memory crash.
-Turning on meta files for external VCS out of memory crash (same on 3.4.2 and lower).
-Switching between complex scenes (especialy after play) = our of memory crash (new on 3.5)

There is HUGE problem related to work with big project (lots of files).

-Frustrum culling does not work properly, for example looking to the ground in the middle of the city makes 6k drawcalls (since unity 3.0), looking on the empty terrain from the city makes same drawcalls like looking on the whole city, etc....tested on other engine and NO problem there.

-occlusion culling finally does not crash during baking, but it does not work properly. Even with very small cubes it does not occlude almost anything (no matter what setup). For example looking to the big wall still renders whole city behind wall (occlussion areas seems precounted OK - before wall, after wall).

[BigBulle]

Hi,
Another possible alternative is to give access to svn ( after prior arrangement with Unity ), for huge projects.
Thing is it is not possible to spot errors without the source. It takes time, is really boring but eventually helps you and rest of Unity community.
-Ippokratis

I'm not sure to understand what you propose. Is it to move Unity to a free open source project?
Personally I start to be really tired to create bug reports for version 3.5 and/or version 3.4.2 to prove this or that to the unity support.

I would really prefer to pay more and to get a stable product. Maybe we have chosen the wrong product for this.

[Ippokratis]

Hi,

I'm not sure to understand what you propose. Is it to move Unity to a free open source project?

Not actually. I mention another possibility of sharing the project, with Unity ( not everybody ), after a prior arrangement with them, which might be helpful in huge projects. I.e imagine you have a 5Gb project you wish to send to Unity. In this case, the above scenario makes it easier. This approach was mentioned by Metron in another thread.
I hope it is more clear now,
-Ippokratis.

[jlevel]

Hi,

Not actually. I mention another possibility of sharing the project, with Unity ( not everybody ), after a prior arrangement with them, which might be helpful in huge projects. I.e imagine you have a 5Gb project you wish to send to Unity. In this case, the above scenario makes it easier. This approach was mentioned by Metron in another thread.
I hope it is more clear now,
-Ippokratis.

I understand your explanation, but again, my work isn't being a beta-tester of a product which I paid more than $2000 before and it's closed-source.

I'm the client which I adquired this product, I'm not the engineer developing it. Maybe sounds a little hard, but it's the reality.

Do you really want improvements from users community? Ok, then make it free and open-source. But don't expect this behaviour will be on a commercial and closed-source product.

I'm tired of send bug reports, some of they are ignored because I didn't attach the full project. Unacceptable.

I have to report a bug, but wait, I can't query the bug tracker, maybe I'm just losing time reporting because I'm just creating a duplicate case. Unacceptable.

If I decided spend this money it's because I trusted on a robust product, what's happened is they released a new "stable" version when really is a beta version.

Furthermore, this "stable" version doesn't allow regression to your project because they are converting the full project to 3.5.

Really, not allowing regression to a previous version, in technology terms, is unacceptable and a suicide.

Errors caused by memory leaks on a commercial product is unacceptable too, and should be fixed as soon as posible with highest priority.

This last maybe is the most important and more critical.

[Ippokratis]

Hi,
jlevel :

I assume that you are responding to Unity, not to me personally. I am a client, much like you. I choose to have a proactive approach and it is clear that everybody makes its choices as he/she wishes. Error reporting is beneficial for me as a customer, so I endorse and suggest it to others. To me, the ability to send error reports with no additional cost is an advantage.

I sincerely sympathize you and anybody who have spend money, time and passion on a project. Before I purchased Unity I made a simple question to myself : What I get now, worth it ? Not what this tool might evolve to. What I get now. So I evaluated the product ( Iphone Basic, after owning Unity Basic ) again and again. At the end, the answer was yes, so I bought it. I do not suggest that everyone should act in the same way, people are different and differences make life interesting. But you can try the product, you can see if it fits your needs, before you buy it.

Another thing is, if someone has a vision and believes that Unity cannot support it, why stay ? Why waiting until this game engine evolves ? Unity costs little - 2000$ are peanuts compared to what costs to produce a game. It is not so difficult to migrate to another game engine or create a new one.

Also this update is free. I am not suggesting that since it is free it is ok not to be "robust" as you say. But it is free. You have not paid for it. You can keep on developing on a previous robust version if you wish and keep a backup of your project ( I found out the hard way too, like you ).

In any case, I just wanted to share my point of view. I really cannot "get in your shoes". I understand that from your point of view, I might be wrong. It is ok. I just hope that the issues with the current update will resolve in a nice way.

Kind regards,
-Ippokratis

[jlevel]

Hi,
jlevel :

I assume that you are responding to Unity, not to me personally. I am a client, much like you. I choose to have a proactive approach and it is clear that everybody makes its choices as he/she wishes. Error reporting is beneficial for me as a customer, so I endorse and suggest it to others. To me, the ability to send error reports with no additional cost is an advantage.

I sincerely sympathize you and anybody who have spend money, time and passion on a project. Before I purchased Unity I made a simple question to myself : What I get now, worth it ? Not what this tool might evolve to. What I get now. So I evaluated the product ( Iphone Basic, after owning Unity Basic ) again and again. At the end, the answer was yes, so I bought it. I do not suggest that everyone should act in the same way, people are different and differences make life interesting. But you can try the product, you can see if it fits your needs, before you buy it.

Another thing is, if someone has a vision and believes that Unity cannot support it, why stay ? Why waiting until this game engine evolves ? Unity costs little - 2000$ are peanuts compared to what costs to produce a game. It is not so difficult to migrate to another game engine or create a new one.

Also this update is free. I am not suggesting that since it is free it is ok not to be "robust" as you say. But it is free. You have not paid for it. You can keep on developing on a previous robust version if you wish and keep a backup of your project ( I found out the hard way too, like you ).

In any case, I just wanted to share my point of view. I really cannot "get in your shoes". I understand that from your point of view, I might be wrong. It is ok. I just hope that the issues with the current update will resolve in a nice way.

Kind regards,
-Ippokratis

Hello Ippokratis,

of course I'm responding to Unity, nothing personal here.

On commercial product is a must to provide support. Yes, 3.x update is free, but what happen if you buy Unity 3.51 Pro? You can't sell something and if it's broken you can't wash your hands. That makes no sense.

Without Pro you can't test occlussion culling, and it's one of the broken things. When you really see this issues it's on a big project, and Pro demo version has expired.

Make a 3D engine from scratch only a standalone developer?? I'm sorry, I'm not superman. If develop a game for an indie takes a lot of time and it's hard, then add this handicap. I don't think so, specially if you target mobile market.

Unity have done a great job in overall, to improve it there is a need to criticism.

Anyway, I'm sure Unity will fix these things, I'm not saying the opposite, just I'm saying what I think they need to improve urgently.

[l0cke]

I understand your explanation, but again, my work isn't being a beta-tester of a product which I paid more than $2000 before and it's closed-source.

I'm the client which I adquired this product, I'm not the engineer developing it. Maybe sounds a little hard, but it's the reality.

Do you really want improvements from users community? Ok, then make it free and open-source. But don't expect this behaviour will be on a commercial and closed-source product.

I'm tired of send bug reports, some of they are ignored because I didn't attach the full project. Unacceptable.

I have to report a bug, but wait, I can't query the bug tracker, maybe I'm just losing time reporting because I'm just creating a duplicate case. Unacceptable.

If I decided spend this money it's because I trusted on a robust product, what's happened is they released a new "stable" version when really is a beta version.

Furthermore, this "stable" version doesn't allow regression to your project because they are converting the full project to 3.5.

Really, not allowing regression to a previous version, in technology terms, is unacceptable and a suicide.

Errors caused by memory leaks on a commercial product is unacceptable too, and should be fixed as soon as posible with highest priority.

This last maybe is the most important and more critical.

Exactly. I agree with each point. Our team has same experience and we have some opinion.

[_rem]

Hello,
I decided to create a new project and re-import all my "assets" one by one. After several hours fighting with broken links, "levels", "Tags", hardlinks between scripts and "gameobject", scripts and scripts, remake "ligthmaps" ... for short:
- Everything works perfectly, no "dracalls" crazy, no display problem with "the prefab Nighttime Water", great! the same performance as version 3.42.
I decided to apply "Occlusion Culling" on my "terrain" and on elements of decor, I check with "visualization" everything is OK, I "Bake".
And there, all problems begin! "dracalls" crazy, refraction of "prefab Nighttime Water" blinks, the framerate collapses ...
I delete data "Occlusion Culling" and "Occlusion Area" and everything returns to normal, I create a "webplayer" the same performance as version 3.42.
- "Occlusion Culling" not work with 3.51 (it did not work, also with 3.42 but silently)
I'll edit my bug report with a simplified project, (I found another strange big bug not easy to reproduce!)
thank you for the support.

Joachim Ante, aNTeNNa trEE : not comment on my observations ? but may be you already knew that "occlusion culling" does not work and a performance penalty?
Otherwise have your views on this issue, before I volunteered to participate your Debug application(working on a simplified project with the bug report) will be welcome !
cordially

[angel_m]

I think Unity needs to stop adding new features to the engine and fix the basis.
I understand the whole engine system is very complex and with the big changes introduced in 3.5 version, the priority must be consolidate it as a robust product.

[BigBulle]

+1000