[sidenimjay]

i think aitraaz has it pretty close . . . .


biased means essentially there are no shortcuts taken......no photon maps to "bias" towards or from, no irradiance caches to "bias" towards or from . . . .

i seriously doubt that a biased render is one that sits back and says i dont think i want to render that, i will render this, but no not that caustic, oh and i will render this a little brighter here cause thats what i think . . . .

this is not ai . . . . unbiased in this case is not a matter of opinion, likes or dislikes, wills and wonts, dos and donts, although it seems to me that is what more than one person in the world seems to think

however the comment above is quite biased . . .

i think houdini is the most supreme animation package out there. Biased statement
maya and houdini have great tools for animating . unbiased statement

this is not the type of unbiased that maxwell claims to be . . . .its a strange little homonym ?

or is it a synonym . . . . damned english . . .

[deesee]

Hey all,

I missed this thread and ivox3 pointed out. I really don't know enough about biased and/or unbiased, but I thought I'd quote one of my one posts to see if having more info about the scene adds to this already spirited discussion. The bottom line is I did have to add two filler planes (one to the right and one to the left as you look at the pic) because it was too dark. Then again maybe it was the fstop and s/s and ISO. I don't know enough about photography.

Having said all that, the time of day in the pic is December 12, at 1200. Now, I live in this apartment and it in fact faces north-north-east, not South as in my image, so Rivoli is right when he sees I made it how I wanted it. However, in June, for a few hours when the sun rises, I get about the most direct sun possible into the room. It is not as bright as the render here, but its fairly close, even with only a little direct light making its way in. If I faced the model according to its true axial direction, and used the actual sun setting, the image is really dark.

The point is, I don't know jack about biased/unbiased, all I know is I had to turn the model and use fillers to get this image. In the end, I don't mind because I like it and maxwell it is easy for me to use.

Well, I revisited this scene one more time and took some more whacks at it. I changed the angle because frankly I just got sick of looking at the other one. If somehow I could get this noisefree....alas.

Stats:

- Sun + skylight
- 2 emiiter planes as fillers (125W each)
- 4 emiiters above counter (35W each)
- 3 emiiters near microwave (10W each)
- 2 vases above cabinets are dielectrics (abbe 200, brownish color)
- glass on coffe table is from glass test on this forum
- floor is plastic (UV 0.2, 0.2; bump 30 w/ specular map)
- shades are diffuse with SSS (abs 0.9, scat 0.2)
- camera fstop 4.0, ss 250 (ISO 400)

RENDER TIME 51 H 18 MIN; SL 20.89

take care all,

deesee

[iandavis]

jandavis instead of repeating your confused an poor lectures on the issue u better read well what i wrote.. i was taking of a room with shutters more than half down and lightened just from indirect light at noon
...u are right ....a so called unbiased render that dont let the light pass trhough glasses is quite far from beeing real
...... anyway everybody is free to bealive to flying horses!!

ehh... whatever.

I have a university degree, been a photographer for twenty five years, and worked in CG since 1989... I don't usually peddle flying horses, unless that is, they are the CG kind.

I have paid my dues, which obligates me to help those who are just starting out. If you feel what I say is bothersome, then obviously I'm wasting my effort.

you wont hear a peep from me again.

cheers.

[Thomas An.]

Hi,

Iandavis is correct. Biased / Unbiased in the rendering context refers to the level of ray randomization used by the internal calculations.

I believe there is confusion between the terms biased/Unbiased and Physically accurate.

Technically, it should be possible to build a renderer that draws cartoons and is also unbiased; meaning that (for example) the pixel at location 301x285 will always have the exact same RGB value no matter how many times we hit th render button.

What giacob ment to say all along relates to the "Physically accurate" aspect.

[lwan]

i think giacob is flaming the renderer because flaming NL don't work anymore

btw i think iandavis is right too. as for "physically accurate", it does not mean a lot of things since science is continuously evolving. some years ago a renderer with inverse square law applied to its lights was assumed physically accurate

[deesee]

If you feel what I say is bothersome, then obviously I'm wasting my effort.

you wont hear a peep from me again.

cheers.

Ian, say it ain't so! I'd hate to see you run off. You helped me out and newbies like me appreciate it. Pull up a chair...stay a while.

deesee

[Frances]

Ian, say it ain't so! I'd hate to see you run off. You helped me out and newbies like me appreciate it. Pull up a chair...stay a while.

I think Ian was speaking specifically about giacob not hearing from him.

[juan]

Hi,

there are two distinct claims made for Maxwell- physically accurate and unbiased

Exactly. Maxwell is unbiased and it means it always converges to the solution without introducing numerical tricks or weird artifacts. Maxwell always garantee the convergence. In the other side Maxwell is very physically accurate due to the internal models that it handles. Saying that Maxwell is not unbiased because there is a bug concerning the light through glasses in some scenes is a bit biased statement however.

Physically accuraccy is a different thing. It is always limited for humans, dogs even insects. We have to recognize that we do not use 5 millions of decimal numbers for PI and also we do not handle each quark separately. But the physical models that Maxwell uses are so accurate than the final result is very close to the reality. Of course this physical models are going to be improved in the future, for example adding a more complex sky model.

Maybe they can optimize the engine to the point where it is fast enough....

Of course. This accuracy takes some times long render times, but, as we have said already, first we develop the most accurate and stable engine and then we will optimize it. There will we continuous enhancements in the speed with each update and you will see amazing render times in the future.

Regards,

Juan

[macray]

Yepp. Finally a statement for us to look forward!!

[ludi]

hey buffos.....did you read that....all becomes good
thx juan

[giacob]

Hi,

there are two distinct claims made for Maxwell- physically accurate and unbiased

Exactly. Maxwell is unbiased and it means it always converges to the solution without introducing numerical tricks or weird artifacts. Maxwell always garantee the convergence. In the other side Maxwell is very physically accurate due to the internal models that it handles. Saying that Maxwell is not unbiased because there is a bug concerning the light through glasses in some scenes is a bit biased statement however.

Physically accuraccy is a different thing. It is always limited for humans, dogs even insects. We have to recognize that we do not use 5 millions of decimal numbers for PI and also we do not handle each quark separately. But the physical models that Maxwell uses are so accurate than the final result is very close to the reality. Of course this physical models are going to be improved in the future, for example adding a more complex sky model.

Maybe they can optimize the engine to the point where it is fast enough....

Of course. This accuracy takes some times long render times, but, as we have said already, first we develop the most accurate and stable engine and then we will optimize it. There will we continuous enhancements in the speed with each update and you will see amazing render times in the future.
Regards,
Juan

juan what do u mean with "Maxwell always garantee the convergence." u mean that if i make the render of a scene with , for instance, mental ray two times with the same identical setting i dont have the same results and with maxwell yes? i cant get what u mean
whatever it means i dont care that much about that. i would greatly prefer that maxwell would let the light enter in my interiors trhough glasses as it happend in the real world !!! a render that dont let the light trough dielettrics in is not at all phisically correct

[giacob]

Hey all,

I missed this thread and ivox3 pointed out. I really don't know enough about biased and/or unbiased, but I thought I'd quote one of my one posts to see if having more info about the scene adds to this already spirited discussion. The bottom line is I did have to add two filler planes (one to the right and one to the left as you look at the pic) because it was too dark. Then again maybe it was the fstop and s/s and ISO. I don't know enough about photography.

Having said all that, the time of day in the pic is December 12, at 1200. Now, I live in this apartment and it in fact faces north-north-east, not South as in my image, so Rivoli is right when he sees I made it how I wanted it. However, in June, for a few hours when the sun rises, I get about the most direct sun possible into the room. It is not as bright as the render here, but its fairly close, even with only a little direct light making its way in. If I faced the model according to its true axial direction, and used the actual sun setting, the image is really dark.

The point is, I don't know jack about biased/unbiased, all I know is I had to turn the model and use fillers to get this image. In the end, I don't mind because I like it and maxwell it is easy for me to use.

Well, I revisited this scene one more time and took some more whacks at it. I changed the angle because frankly I just got sick of looking at the other one. If somehow I could get this noisefree....alas.

Stats:

- Sun + skylight
- 2 emiiter planes as fillers (125W each)
- 4 emiiters above counter (35W each)
- 3 emiiters near microwave (10W each)
- 2 vases above cabinets are dielectrics (abbe 200, brownish color)
- glass on coffe table is from glass test on this forum
- floor is plastic (UV 0.2, 0.2; bump 30 w/ specular map)
- shades are diffuse with SSS (abs 0.9, scat 0.2)
- camera fstop 4.0, ss 250 (ISO 400)

RENDER TIME 51 H 18 MIN; SL 20.89

take care all,

deesee

i dont think that if u would take a picture at 12,00 in a room with such a large window u would get such a dark picture which seems to be shot shortly before sunset

[vkiuru]

i dont think that if u would take a picture at 12,00 in a room with such a large window u would get such a dark picture which seems to be shot shortly before sunset

Would depend on how your camera was set up.

http://www.mortalcoil.com/pix/1999/07-J ... 20room.jpg
http://www.myimpactengine.com/members/0 ... 20Room.JPG
http://www.newyork-hotels-classify.com/ ... nt/rm1.jpg

Or am I missing the point?

Hehe- that is definitely not a noon sun angle. I don't think it has anything to do with biased/unbiased, etc... it just seems that the math that positions the sun isn't working correctly yet.

Depends on the month and latitude/longitude.. deesee said the time in the pic was December 12 at noon. Have a look at a January, middle day in Finland:
http://kotisivu.dnainternet.net/jopejahu/ppa/ppa011.htm

[Mihai]

giacob....how long are you going to drag this on?

You are wrong and you refuse to accept some facts. Move on.

[giacob]

giacob....how long are you going to drag this on?

You are wrong and you refuse to accept some facts. Move on.

yes when u will biodegrade yourself off (100% though) i will stop