FakeaoCPU not working in 32 bit v17?

[NancyGormezano]

I get a crash everytime - with fakeaoCPU with ver17 32bit (most recent release), even with a simple chor (1 sphere)

 

Anyone else getting it to work? Is it just me?

 

EDIT: I've added a test project (created in 16b) that crashes in 17.

 

I tried creating project from scratch in 17 - still crashes.

fakeao16b.prj

Edited August 7, 2012 by NancyGormezano

[John Bigboote]

I'm DL'ing the new 17R... will try shortly.

 

EDIT: Wull, it worked. (a peacock image is missing) The ONE thing I may have done differently, is I look to see if you had the buffers set right (depth, color) and you did... and then I always look to see if the FakeAO's camera's focal length setting is the same as the REAL camera's setting- and I saw you had it cranked-up to 250. So, I set it back to 70(your cameras focal length) and rendered. The WEIRD thing... is that after I had rendered, uploaded to this post- and then went to close the A:M render screen- I crashed. I'm looking in to see if that repeats. EDIT to EDIT: 2nd time around it did not crash... man alive- this new release is greased-lightning in the RENDER department!!! Whoooo!

[NancyGormezano]

I'm DL'ing the new 17R... will try shortly.

 

EDIT: Wull, it worked. (a peacock image is missing) The ONE thing I may have done differently, is I look to see if you had the buffers set right (depth, color) and you did... and then I always look to see if the FakeAO's camera's focal length setting is the same as the REAL camera's setting- and I saw you had it cranked-up to 250. So, I set it back to 70(your cameras focal length) and rendered. The WEIRD thing... is that after I had rendered, uploaded to this post- and then went to close the A:M render screen- I crashed. I'm looking in to see if that repeats. EDIT to EDIT: 2nd time around it did not crash... man alive- this new release is greased-lightning in the RENDER department!!! Whoooo!

 

Yes I usually have my Focal length cranked to 250 (makes the effect darker for some reason)

 

Did you test with 32 bit A:M? or 64bit? And I do not have the GPU version, only CPU version of FakeAO. I am 32bit.

 

(weird about the peacock image - I don't have that crop up in either 16b or 17)

 

EDIT: just looked at text file for project - and yes the reference for the peacock image is there, but it does not bring it into the project for me in either 16b or 17. I had used that image in testing at one time for testing Image based lighting with an image. Seems to be leftover garbage.

[John Bigboote]

I have both, but used what you had it set at which was FastAO CPU version... my test was on the 64 bit version of the new A:M Version 17.0 SSE4

 

I remember when we were all new to FastAO, cranking-up the focal length was a 'trick' to get an enhanced effect (I think it was the trial version and we were trying to push it further.) This is not necessary anymore, you can increase the gamma setting higher than 1 (I like 1.3ish) and you can increase the SamplingLevel and # of Samples ever higher now.

 

I did a movie with a simple camera move... and cranked up settings (but NOT the focal length) the ONLY thing I do not like is the way the AO texture travels with the scene... something unnatural about it...

Nancy_000.mov

[NancyGormezano]

my test was on the 64 bit version of the new A:M Version 17.0 SSE4

 

I did a movie with a simple camera move... and cranked up settings (but NOT the focal length) the ONLY thing I do not like is the way the AO texture travels with the scene... something unnatural about it...

 

Can you try it in the 32 bit version to see if you crash? I do not have 64bit. Nor do I have SSE4 (I don't think...how would I know?)

 

Yes that aotexture moving does look weird - but I never noticed that in 16b. Could be an artifact of FAKEAO since it is based on screen space depth, and not real raytracing type. Probably never noticed it in my animations because I texture the sh..t outta things

 

I don't like to increase sampling levels, #samples too much, makes things too slow on my machine. But yes, I can increase the gamma, as well as focal length.

 

EDIT: I'm guessing you have waaaay more memory than me, and are using win7. Perhaps that is why there was a delayed crash for you? memory leak maybe? (me: 4GB - but only 2GB available to programs?, winxp pro sp3)

Edited August 7, 2012 by NancyGormezano

[John Bigboote]

Hey Nancy- the SSE4 I got from the 'Help' and 'About Animation Master' splash...

 

I am stumped with V17.0/32 and FakeAO. A:M does not recognize that I have it upon booting, and will not see it as an option under New/PostEffect. My Version 17/64 and 17/32 are both installed to the same directory so they both read from the same plug-in folders... I gotta figger this out- can't do a test for you right now... are you still crashing?

[NancyGormezano]

I gotta figger this out- can't do a test for you right now... are you still crashing?

 

yes I still crash. No matter what settings (default) I try with Fakaocpu.

[John Bigboote]

Yeah, me too. I didn't have my 32 bit FakeAO thingy installed... so I found it... installed it... tried it... and crashed. I even set the Focal Length to 70. I can verify your dilemma!

 

Yup- deleted your post effect and made a new one of my own... still crash.

[NancyGormezano]

Yeah, me too. I didn't have my 32 bit FakeAO thingy installed... so I found it... installed it... tried it... and crashed. I even set the Focal Length to 70. I can verify your dilemma!

 

Yup- deleted your post effect and made a new one of my own... still crash.

 

Thanks Matt - That is very helpful to narrow this down to 32bit version!

[robcat2075]

I get a crash also in 32-bit A:M.

 

AMReport it and note that it works OK in v16.

[John Bigboote]

AMReport it and note that it works OK in v16.

 

And 17.0/64bit. Do you report to A:M or Jenpy?

[NancyGormezano]

I get a crash also in 32-bit A:M.

 

AMReport it and note that it works OK in v16.

 

I had already done that earlier today. Was checking to see if we could narrow it down to 32bit. Thanks for testing.

[robcat2075]

I just tried it in v17 64-bit and it does work. Mention that too.

[yoda64]

Okay guy's , I was diving today into this problem .

Bad news , the plugin will work only with v17 , when jenpy recompile the plugins with the new v17 sdk (avaible here)

 

The problem is coming from a internally change (affecting the Time class, which is also used from the sdk)

/* from the changelog */

!Breaking changes

- MAX_TICKS changed from long to long long to avoid overflow the long,

when squaring time (issue #0005893: Expression rotation limited to 16,000 degrees)

- Baked particle systems

particle systems baked with version 16 or previous, won't

work in V17 , they need to be rebaked (coming from the above change)

/* */

 

Changing this back isn't a option .

[robcat2075]

Okay guy's , I was diving today into this problem .

Bad news , the plugin will work only with v17 , when jenpy recompile the plugins with the new v17 sdk (avaible here)...

 

 

Someone turn on the jenpy-light!

[NancyGormezano]

ugh. I am confused: It works in 64 bit version but not in 32? Was it recompiled for the 64 bit version and not the 32 bit version?

[robcat2075]

ugh. I am confused: It works in 64 bit version but not in 32? Was it recompiled for the 64 bit version and not the 32 bit version?

 

The plugins (one for 32 and one for 64-bit) haven't been recompiled.

 

I guess it's just crazy C++ stuff that keeps the 64-bit version working in spite of the change.