PDA

View Full Version : The Shining - elevator of blood



Mr Rid
04-11-2010, 02:52 AM
http://www.youtube.com/watch?v=7049WVfSanM Select 720p and full screen.

This is a half-scale, CG sim that I did just for amusement in spare hours. I never expected it to look exactly like the real thing or to have taken it even this far. The RealFlow fluid sim uses only about 1.6 million particles and so the resulting meshes appear thicker and 'blobbyer' than an actual water-like liquid at this scale. I think at least ten million particles would begin to look convincing, but since this took about a month to calculate and render on my I7, 3.2gh quad core... :eek:

I rendered and made simple models of the hallway and furniture in LW 9.6, and post processing was applied in Fusion 6 in attempt to more resemble the original film. Many of the tiny droplets are voxels that were also thrown in.

To come up with the dimensions of the set, I studied each shot in The Shining of the various hallways and fixtures. I settled on a 50mm lens, and my guess is that it was a half scale miniature since Kubrick was a hardcore perfectionist and smaller scale liquids look progressively less realistic and blobby (the scale was also more RealFlow friendly).

The angles of the walls and lines on the floor indicated that the camera is not centered but was placed a few feet to the right and panned left to appear centered. Most viewers do not notice there are actually 3 different angles of this scene shown in the movie. This leads me to suspect there may have been another camera centered on the miniature set but the footage was not used, since that angle does not otherwise appear in the film.

I arrived at a 140fps frame speed by using Fusion to speed up the slow motion shot from the film until it appeared to move at a realistic speed which was x6. 6 x 24 = 140. To save time, I began RealFlow simulations at 24fps. When I sped up the resulting render x6, the CG blood turned out to hit the walls and run out of frame on close to the same frames as in the real shot, so this affirmed the rate of 140fps and the dimensions of the set as being somewhat accurate. When they shot the miniature scene, the actual event would have taken place in about four seconds. The final 140fps sim and render total 535 frames.

The overhead render was an afterthought where more flaws are apparent.

I started thinking about rendering this after viewing a YouTube video titled 'THE SHINING - SOMETHING IN THE RIVER OF BLOOD' where the author presents a lengthy assertion on some obscure shapes viewable within the blood near the elevator that Kubrick intended as a sublminal representation of a corpse or something suggesting 'Tony' inside Danny's mouth. The seemingly mysterious shapes present a Rorschach image for commentors' imaginations to offer up a myriad of odd things they believe 'it' to be. It was apparent to me that the shapes are merely reflections of the set and I emailed the author a simple CG still demonstrating that fact but his ego defended the presence of mysterious objects.

But I thought it would be interesting to see what ReaFlow might do, and was surprised to find that no one else had attempted a Shining-blood-elevator sim before that I could find. Oddly enough, very similar 'shapes' turn up in this CG render as in the movie since they are simply distorted reflections of the elevator door itself and surrounding frame and wall. But hey, people prefer a mystery.

Cageman
04-11-2010, 03:02 AM
Absolutely FANTASTIC!

Talk about inspirational! Well done! :thumbsup:

Iain
04-11-2010, 03:23 AM
Great job! Very convincing reconstruction.

That scene really freaked me out as a kid.
Did it really take a month to calculate?

Lewis
04-11-2010, 03:34 AM
Wow, it looks extremely good, but man 30 days to calculate on i7 system ? Damn that's slow calculation :). Can it calculate (realFlow) over Renderfarm/network ?

Mr Rid
04-11-2010, 04:04 AM
...
Did it really take a month to calculate?

Over a couple of months I spent a few hours most days trying different sims in RF and letting them run overnight. You have to experiment a lot in RF, and I had the biggest problem coming up with proper collision geometry. I also never got RF's damn wet maps to work right and had to post edit them extensively.

When I first filled the elevator with 1.6 mil particles, it took a week for them to settle and stop sloshing around. It then took a week to calculate the 535 frame spill, the resulting particle bin files taking up 162gb. In LW, the 720x480 frames averaged 20 minutes a frame which took another week, eating up all 12gb of ram at the peak and going about 4gb into virtual. Then another week to render the overhead shot. More days were spent re-rendering some problem areas and then working out the comp.


...Can it calculate (realFlow) over Renderfarm/network ?
Unfortunatley they have never figured out a way to distribute sim calculation over a network.

Cageman
04-11-2010, 04:27 AM
I've been watching this over and over again and I really like the very, very subtle camera shakes that you have added in there (or did you do that in post maybe?).

It helps selling the realism so much. :)

Mr Rid
04-11-2010, 05:19 AM
I've been watching this over and over again and I really like the very, very subtle camera shakes that you have added in there (or did you do that in post maybe?).

It helps selling the realism so much. :)
ReaFlow sims can be mesmerizing. :)

Projected film always wiggles or 'orbits' a little due to sprocket wear and the mechanical nature of a projector, even when converted to or from digital. Movie theater projectors are very worn and the film wriggles and streaks vertically (especially noticeable with credits over black) a great deal, but audiences are used to it and tend not to notice. I once worked as a projectionist and am always amazed that theater projectors and the prints hold up as well as they do. If any tiny part of the complex mechanism is off a millimeter the film can easily scratch, tear and break (which happened regularly and I had to quick splice). Of course digital projectors are now rock solid.

In film FX, we may spend time matching colors and tracking in a film plate, then later they decide to extend the shot and we have to get a new version of the plate. But no two scanned plates of the same shot are ever exactly the same. Each is positioned slightly different in frame, there are often changes in color, and the wiggle has a different rhythm so you have to re-do the comp and tracking. Thats film. Its all the more amazing looking at old printer FX like Star Wars where they had to simultaneously synch many pieces of film, all wiggling differently on every pass.

But I use a separate LW scene with the sequence mapped on a card to add shake. I just know how to get it the way I want in LW from old school as it never looks right in Fusion or AE. I also added a faint vignette as older films tend to appear brighter in the middle where the lamp is. It doesnt come across much on YT but I also added two types of film grain, slight dirt and defocus.

nikfaulkner
04-11-2010, 05:21 AM
that looks great,

funny, that was one of the first things i wanted to try to reacreate when i first tried blender fluids.

very well done

n.

Cageman
04-11-2010, 05:29 AM
Cool! :)

Did you resort to any type of breakup for the rendering? Or is all in camera?

Elmar Moelzer
04-11-2010, 09:19 AM
Awesome work man!
Also love the reel. Did not know you used to work at UFO. They always were my heroes back in the good ol' days ;)

Rabbitpenny
04-11-2010, 09:21 AM
Holy cow! Impressive as usual, Mr. Rid. Now, there is no mystery why a simulation has never been done before.

shrox
04-11-2010, 09:27 AM
Based on the times described by Mr. Rid, I would say he does not have ADD. Remarkably skilled yes, ADD no.

shrox
04-11-2010, 10:43 AM
Do the wet maps simulate the residue left by splashes? Like on the wall where the blood gushed up against it?

robyht
04-11-2010, 11:05 AM
stunning! go LW! and Realflow and Fusion of course

OnlineRender
04-11-2010, 11:10 AM
excellent ,amazing work as per ! now do it with blender fluids :)
--------------
edit you forgot the hidden unidentifed object :P

geothefaust
04-11-2010, 11:47 AM
Looking damn sharp Mr Rid! It looks amazing, and close to what I remember. Out of sheer curiosity I'll have to play that scene from the movie and compare.

Red_Oddity
04-11-2010, 12:24 PM
Over a couple of months I spent a few hours most days trying different sims in RF and letting them run overnight. You have to experiment a lot in RF, and I had the biggest problem coming up with proper collision geometry. I also never got RF's damn wet maps to work right and had to post edit them extensively.

When I first filled the elevator with 1.6 mil particles, it took a week for them to settle and stop sloshing around. It then took a week to calculate the 535 frame spill, the resulting particle bin files taking up 162gb. In LW, the 720x480 frames averaged 20 minutes a frame which took another week, eating up all 12gb of ram at the peak and going about 4gb into virtual. Then another week to render the overhead shot. More days were spent re-rendering some problem areas and then working out the comp.


Unfortunately they have never figured out a way to distribute sim calculation over a network.

Really good looking sim there Mr Rid.

Have you tried settling your particles with a kSpeed daemon or by resetting their velocity with an Event script? That usually speeds up settling by a good amount. (oh, and make sure velocity display is not set to auto, this normalizes the velocity display so you'll keep seeing white high velocity particles, even though they only have a tiny velocity.)

And, yeah, large bin files can become a problem when you create a lot of variations, last project we did took almost 700GB of storage on the bin files alone.
Luckily as soon as a sim has been green lit for rendering, you can dump all the other bin files you don't need.

I can't wait for RF5 with its new hybrid solver and improved particle based solver (supposedly it can be up to 20 times faster with sparse distribution sims, and they got rid of the annoying licensing system they have now with RF4)

Mr Rid
04-11-2010, 01:15 PM
Now that digital is here, I'm going to miss grunging up CG to make it look like film.



Did you resort to any type of breakup for the rendering? Or is all in camera?

I 'resorted' to rendering a separate matte for the blood in order to post tweak the color slightly. And I had to render some parts of the floor without blood or wetmap and then animate a mask between the two versions in comp since the wetmaps insisted on putting splotches where there were no particle collisions. I refashioned the hallway and UV several times but it just refused to work much to my annoyance (as usual, I am the only entity on the planet to have this problem apparently). I am guessing I would have to break out each wall and floor into separate pieces with simple planar UVs for it to work. An early screwy wetmap- 83926


Do the wet maps simulate the residue left by splashes? Like on the wall where the blood gushed up against it?
I never tried RF wetmaps before, but yes they make corresponding splotches that may blur and fade over time. http://vimeo.com/2679536

DP has a surfacing thing that works similarly with PFX colisions that I have never tried. RF sims are now suppose to be able to convert to PFX.

I also have an old work around for this in LW that I have used for wet maps or a wipe effect like a brush painting a surface. Its done by saving out the specific shadow alpha isolating where ever the liquid (or other geom) intersects with geometry. Then the alpha sequence is added together with itself in a separate LW scene by loading into the background and also mapping onto a foreground card set to additive and with a 1 frame offset. This is a little weird to wrap the brain around at first, but the sequence that LW is saving out is simultaneously set to load and map on the card. The additive combines the two alphas as the next saved frame is superimposed over the last frame resulting in a trail effect.
83925

The mipmap strength can blur the image which will add to each subsequent frame and slowly increase blur over time.

Mr Rid
04-11-2010, 02:16 PM
R
Have you tried settling your particles with a kSpeed daemon or by resetting their velocity with an Event script?

No. Forget it if I ever have to type or figure a script. I cant figure out what the hell you type in the dos window to run RF without the interface either, but I dont imagine that makes much speed difference with this. I started the sim with higher viscosity and collision friction to make it more sluggish and settle sooner, but am not sure how much difference it made. I saved out bins during the settling instead of locking the flow, so I could preview what was going on and see how close it was to settling. Otherwise it may appear relatively level on a given frame when it may actually just be between sloshes.

I would also rather use the meshes instead of the bins in LW since dense bins take minutes to interpret and load on each frame as opposed to seconds with the meshes (LW previews took forever, and usually crashed for some reason). But the new plugins only read bins.

Matt
04-11-2010, 05:27 PM
That was cool!

Tippsy
04-11-2010, 05:54 PM
Amazing, so u were saying if it was up to about 10 million then it would be a thinner more blood type liquid?

VonBon
04-11-2010, 07:43 PM
That was cool. you did a tree piece too, also crazy. :bowdown:

adk
04-11-2010, 08:23 PM
That's some seriously kick @rse work there Mr Rid. My hat off to you for your perseverance & thanks a bunch for the in-depth descriptions. I love reading about how such are accomplished :thumbsup::thumbsup::thumbsup:

Shiny_Mike
04-11-2010, 09:53 PM
that is just way too cool, and masterfully executed!

sampei
04-12-2010, 03:25 AM
mr. Rid the quality of your work is simply incredible. Thanks for sharing all the details. :thumbsup:

Mr Rid
04-12-2010, 04:11 AM
Thank you all for the kind notes. Am glad if anyone learns a little something or is inspired to try something new. But the blobby liquid just ruins it for me. Its frustrating that a fast machine bogs on nowhere near enough particles. If RF5 speeds things up I might try it again.

But I want to be able to handle 27 million particles-
http://www.fxguide.com/modules/NewsUpload/files/10Mar/naiad/water_sim3.mov

Avatar's 'Naiad' solver in beta-
http://www.fxguide.com/article606.html

Red_Oddity
04-12-2010, 05:24 AM
Thank you all for the kind notes. Am glad if anyone learns a little something or is inspired to try something new. But the blobby liquid just ruins it for me. Its frustrating that a fast machine bogs on nowhere near enough particles. If RF5 speeds things up I might try it again.

But I want to be able to handle 27 million particles-
http://www.fxguide.com/modules/NewsUpload/files/10Mar/naiad/water_sim3.mov

Avatar's 'Naiad' solver in beta-
http://www.fxguide.com/article606.html

I was curious about that one as well.

I just wish ScanlineVFX's fluid solver wasn't a service but a piece of software i could screw about with.

http://www.scanlinevfx.com/muc/de/reels.html

WilliamVaughan
04-12-2010, 05:32 AM
Brilliant! One of my top 10 movies.:thumbsup:

Just in time for the 30th Anniversary as well.

Fadlabi
04-12-2010, 05:53 AM
very cool!

safetyman
04-12-2010, 06:31 AM
Just incredible, Mr. Rid. We are so fortunate to have professionals like you in this community. Next you should try the wall spray scene from Evil Dead 2!

Mr Rid
04-12-2010, 06:49 AM
I was curious about that one as well.

I just wish ScanlineVFX's fluid solver wasn't a service but a piece of software i could screw about with.


I happen to know someone else developing a complex fluid solver but cant say who. :angel: They want to trial it on my Shining geometry. And no, it wont support LW.

praa
04-14-2010, 10:30 PM
I happen to know someone else developing a complex fluid solver...And no, it wont support LW.

i say you go thru their door with that axe of yours saying something like here's NEWTEK!

Mr Rid
04-14-2010, 10:48 PM
We'll see, maybe if Core ever gets it going.

Mr Rid
04-14-2010, 10:59 PM
Not to waste anyone else's time but if you have nothing better to do, please comment some sense into this guy- http://www.youtube.com/watch?v=t7-GKHM5HZ8&feature=related

Ive been in a little debate with the author Rob Ager about subliminal objects vs reflections in the Shining blood. Most people dont understand anything about lighting and so they project 'Jesus' face on a piece of toast' so to speak. Commentors see everything from the twins to Danny's Big Wheel hidden in the blood.

An interesting thing about this scene is how the blood that shoots up from the floor seems to splash too high to be occurring spontaneously. There seem to be secondary jets of blood coming up from the sides which causes a lot of speculation. But when I set up the CG version based on various views of the hotel halls, I discovered the shape of the corner of the wall would channel the blood upward more forcefully, but the corner can not be seen from the camera's POV. But you can see it in the overhead render. On set they may have even cheated the shape further with a curled ramp near the floor in order to get a more dramatic splash.

adk
04-14-2010, 11:41 PM
Not to waste anyone else's time but if you have nothing better to do, please comment some sense into this guy- http://www.youtube.com/watch?v=t7-GKHM5HZ8&feature=related

Ive been in a little debate with the author Rob Ager about subliminal objects vs reflections in the Shining blood. Most people dont understand anything about lighting and so they project 'Jesus' face on a piece of toast' so to speak. Commentors see everything from the twins to Danny's Big Wheel hidden in the blood.

An interesting thing about this scene is how the blood that shoots up from the floor seems to splash too high to be occurring spontaneously. There seem to be secondary jets of blood coming up from the sides which causes a lot of speculation. But when I set up the CG version based on various views of the hotel halls, I discovered the shape of the corner of the wall would channel the blood upward more forcefully, but the corner can not be seen from the camera's POV. But you can see it in the overhead render. On set they may have even cheated the shape further with a curled ramp near the floor in order to get a more dramatic splash.

Not discounting the brilliance & subliminal aspects of Kubrick but Occam's razor holds up just fine in this case I reckon.

Your follow up explaining the reflections is really well done Mr Rid. Kudos. Got me convinced it's just reflections ... as for the other guy ?? :D

I'm still amazed how your model comes so close to the original.

Red_Oddity
04-15-2010, 04:37 AM
Not to waste anyone else's time but if you have nothing better to do, please comment some sense into this guy- http://www.youtube.com/watch?v=t7-GKHM5HZ8&feature=related

Ive been in a little debate with the author Rob Ager about subliminal objects vs reflections in the Shining blood. Most people dont understand anything about lighting and so they project 'Jesus' face on a piece of toast' so to speak. Commentors see everything from the twins to Danny's Big Wheel hidden in the blood.

An interesting thing about this scene is how the blood that shoots up from the floor seems to splash too high to be occurring spontaneously. There seem to be secondary jets of blood coming up from the sides which causes a lot of speculation. But when I set up the CG version based on various views of the hotel halls, I discovered the shape of the corner of the wall would channel the blood upward more forcefully, but the corner can not be seen from the camera's POV. But you can see it in the overhead render. On set they may have even cheated the shape further with a curled ramp near the floor in order to get a more dramatic splash.

I rather not, if find discussions on Youtube to be even less fun then trying to have a civilized conversation about LW on the CGtalk forums.

But i also think people over-analyze Kubrick movies, sure the guy was a brilliant completely bat crazy tyrannical movie maker, but come on, you know?

Mr Rid
04-15-2010, 04:45 AM
Not discounting the brilliance & subliminal aspects of Kubrick but Occam's razor holds up just fine in this case I reckon.

Your follow up explaining the reflections is really well done Mr Rid. Kudos. Got me convinced it's just reflections ... as for the other guy ?? :D

He seems like a smart fella but he is just way off on this one.



I'm still amazed how your model comes so close to the original.
I think the dimensions are fairly accurate. I realized too late that my lens (50mm) is not quite as long as the real one.

The voxels can be seen as bright red in the shaded view-

84036

erikals
04-15-2010, 05:57 AM
.....
just a small crit, i noticed in your render some of the particles are "turned off" mid-air.
(edit: my mistake, it was flickering in low-res video that made it appear so)

also, had an idea of subdividing each RF obj, for then to add a procedural displacement texture. (to add fake detail)
unfortunately not sure how to do it automatically.
-import .obj objects in Layout
-export .lwo objects from Layout (unless RF exports .lwo natively, can't recall)
-now, what is needed in Modeler is a script to batch open each .lwo, subdivide it, and close it.

must say, Great render! http://erikalstad.com/backup/anims.php_files/smile.gif
i didn't really think about that the room itself was CG

.....

erikals
04-15-2010, 06:18 AM
We'll see, maybe if Core ever gets it going.
any idea on what it will be like? compared to RF that is...

Cageman
04-15-2010, 02:58 PM
-now, what is needed in Modeler is a script to batch open each .lwo, subdivide it, and close it.

Hmm... I'm pretty sure there is a script/plugin somewhere that allows you to enable/disable subpatches in Layout.

Doing that enmasse could quite possibly be achived with Dan Dulberger's ef MultiService with LastCommand Master Plugin (http://people.dsv.su.se/~dan-dulb/evenflow3d/index.php) (works great for many things in LW that natively does not support multi selection, such as Record Pivot Rotation and...well... you get the idea). :D

Cageman
04-15-2010, 03:01 PM
Found that subpatcher tool for Layout... seems that you can do it enMasse as well and it is designed to work with object sequences. :)

http://www.blochi.com/gfx/subpatcher_en.html

EDIT: Love this description:

"To TAB an Object sequence you only need one of the objects loaded. This option is most useful for smoothing out those RealFlow simulations...."

:D

Mr Rid
04-15-2010, 03:52 PM
.....
just a small crit, i noticed in your render some of the particles are "turned off" mid-air.
(edit: my mistake, it was flickering in low-res video that made it appear so)

Most of the smallest particles seen are voxels to break up the blobby meshes a little, and the size is shrinking over time so yeah they start to disappear in the processing. Some small particles are bouncing off the wall undesirably or scooting across the floor like little Hot Wheels that are gladly squelched in the defocus and compression.

Part of the RF mesh is also popping on a couple of frames, wet maps are only applied to the white walls (which was enough hassle), the chair textures are too uniform (a cubic map... I find modeling and texture painting tedious). It was tricky to balance blood surface settings that worked well in thinner areas but not so well in thicker areas. Upon close inspection the RF meshes quiver slightly as the poly flow changes on every frame and is most apparent in spec.



also, had an idea of subdividing each RF obj, for then to add a procedural displacement texture. (to add fake detail)
unfortunately not sure how to do it automatically.
-import .obj objects in Layout
-export .lwo objects from Layout (unless RF exports .lwo natively, can't recall)
-now, what is needed in Modeler is a script to batch open each .lwo, subdivide it, and close it.

.....

RF will output an lwo sequence or bin that LW then converts into lwo but dense bins like these take up to 4 minutes (on a quad) per frame to translate and load. The meshes naturally load much faster but the RF plugin is currently crashing on high res lwo, and the motion blur does not work as well as with the bins.

But either way results in a regular mesh that you may apply displacement to as is. In another project I added displacement for a boiling effect. But like this sequence was getting over 2 million polys and I wouldnt want to slow it down even more with subD.

84058

There is a mesh optimization feature in RF I have not tried lately as it used to always make the mesh sequence too jittery frame to frame.


any idea on what it will be like? compared to RF that is...

Well, they're making fun of the Naiad example if thats any indication.

erikals
04-15-2010, 05:42 PM
Found that subpatcher tool for Layout... seems that you can do it enMasse as well and it is designed to work with object sequences. :)

http://www.blochi.com/gfx/subpatcher_en.html

EDIT: Love this description:

"To TAB an Object sequence you only need one of the objects loaded. This option is most useful for smoothing out those RealFlow simulations...."

:D

Fantastic!!! http://erikalstad.com/backup/anims.php_files/023.gifhttp://erikalstad.com/backup/anims.php_files/023.gif
thank you, you've no idea how long i've been looking for something like that http://erikalstad.com/backup/anims.php_files/smile.gif

erikals
04-15-2010, 07:24 PM
Well, they're making fun of the Naiad example if that's any indication.
Naiad, myea... always thought it was a bit hyped.
but only saw an early demo, who knows...
sounds powerful.
do wonder what RF5 will bring though. and good they finally remove the licensing system limitation.

yeah, maybe going from 2mill to 8mill is a bit much.
i think it can be of use though. esp in blobby renderes.
if i recall correct they did something simillar on poseidon.

btw, can RF put weightmap info on smaller part of the mesh?
could be usefull too.


it's awefully hard to replicate a scene like that though,
so much particles are needed, at times one could wonder if
it would be easier to hand-model/animate the thing. or, just do it the old fashioned way.

Mr Rid
04-15-2010, 07:41 PM
Naiad, myea... always thought it was a bit hyped.
but only saw an early demo, who knows...
sounds powerful.
do wonder what RF5 will bring though. and good they finally remove the licensing system limitation.

yeah, maybe going from 2mill to 8mill is a bit much.
i think it can be of use though. esp in blobby renderes.
if i recall correct they did something simillar on poseidon.

btw, can RF put weightmap info on smaller part of the mesh?
could be usefull too.



I find that everything is overhyped until it is actually working on a tight schedule.

It looked like RF was only used on some smaller sims in a few hallway shots on Poseidon. But the Flowline stuff with the boat flipping over in the wave is way outta RF's current league.

But yes, RF outputs weights, UVs, and all kindsa other useful data if you write a translater (open SDK to do so) like vorticity, temperature, mass, pressure, density, viscosity, isolation, age, neighbors, force, velocity, id. Stuff I will never use until someone slaps a friendly GUI on it.

Like this groovie use of vorticity for white water shading RF in Krakatoa-
http://www.youtube.com/watch?v=m7AKrMmaXSg

shrox
04-15-2010, 07:47 PM
...Stuff I will never use until someone slaps a friendly GUI on it...

I take comfort in knowing I am not the only one...

erikals
04-16-2010, 05:54 AM
But yes, RF outputs weights, UVs, and all kindsa other useful data if you write a translater (open SDK to do so) like vorticity, temperature, mass, pressure, density, viscosity, isolation, age, neighbors, force, velocity, id. Stuff I will never use until someone slaps a friendly GUI on it

found this page, looks to have some cool scripts,
http://www.realflowforum.com/view_forum.php?id=30

also,
http://www.digitaltutors.com/09/training.php?cid=31&pid=3578

and....
http://www.nextlimit.com/nlscript/ranking_scripts.php?id=37

here's a foam script, might be of interest,
http://www.realflowforum.com/view_topic.php?id=5790


(attached, the weightmap idea)

Red_Oddity
04-16-2010, 09:26 AM
I just wish i could buy Smorganic for Realflow from FusionCIS, it's one hell of a tool for creating real looking small scale water sims, as it prevents the typical breaking up blobbiness of thin fluid 'sheets'.

I'm trying to program something similar, but damn, there's alot of stuff to think about when doing a Smorganic like event script.

See the FusionCIS channel for examples on Smorganic : http://www.youtube.com/user/fusioncis#p/u/20/e2k6ZMG0RkQ

erikals
04-16-2010, 09:46 AM
maybe modeling it instead for then to use metanurbs could work...
(pros = no calculation time, cons = hard to model/animate)

OnlineRender
04-17-2010, 03:38 AM
I posted it via my FB channel * break down / response * and even my mates who know nothing about animation , said it was AWESOME !

dballesg
04-17-2010, 06:25 AM
Hi Mr. Rid,

Excellent fluid work you have done. Very impressive.

More about Naiad if you are interested is in Exotic Matter (http://www.exoticmatter.com/) webpage.

Looks like there will be only version for Linux and Mac. No windows (well not currently)??? :D

One of Naiad authors Robert Bridson, wrote a very interesting book "Fluid Simulation for Computer Graphics", that I bought and read a couple of weeks ago.

If you pass off the really complex mathematics that he explain in it, there are a few notes there explaining the grid vs. particles methods, really well explained.

Sadly no source code of a full simulator on it, it's only 227 pages :D

David

Mr Rid
04-18-2010, 02:39 AM
More CG Shining, from the Tron Legacy director-
http://www.youtube.com/watch?v=CunBip6tcMY

m0184you
04-18-2010, 11:33 AM
The Shining. What a movie!
Nice work Mr Rid!

Mr Rid
04-18-2010, 04:58 PM
I just wish i could buy Smorganic for Realflow from FusionCIS, it's one hell of a tool for creating real looking small scale water sims, as it prevents the typical breaking up blobbiness of thin fluid 'sheets'.

I'm trying to program something similar, but damn, there's alot of stuff to think about when doing a Smorganic like event script.

No doubt. Yeah, I dont like the cheese-hole effect in RF. RF always tends to look gloppy like melted ice cream unless you have buzillions of particles which is never practical timewise. But most watery-looking sims are usually just experimental of pouring in a simple box or a glass that is more affordable to calculate. But for FX shots with more complex/time-consuming collision geometry, they usually use little cheats and post work to get it to look convincing.

aidenvfx
04-18-2010, 09:37 PM
No doubt. Yeah, I dont like the cheese-hole effect in RF. RF always tends to look gloppy like melted ice cream unless you have buzillions of particles which is never practical timewise. But most watery-looking sims are usually just experimental of pouring in a simple box or a glass that is more affordable to calculate. But for FX shots with more complex/time-consuming collision geometry, they usually use little cheats and post work to get it to look convincing.

Are you running on a Mac or PC also how much RAM do you have. I am curious as I always seem to run out of RAM when using RF.

Mr Rid
04-18-2010, 09:48 PM
Are you running on a Mac or PC also how much RAM do you have. I am curious as I always seem to run out of RAM when using RF.

I ran this sim on an I7 quad, with 12gb ram. The LW render peaked an additional 4.5gb into virtual, although I dont understand why.

Infinite
04-19-2010, 04:03 AM
That was incredible, the fact that the 24fps version makes it look ilke a real life scale model is amazing. The attention to detail here is mind blowing. Well done.

An iconic film and a superb use of Lighwtave and real flow.

Newtek! a great Gallery submission here!

PS it reminded me of this short by Joseph Kosinski - http://www.youtube.com/watch?v=CunBip6tcMY There is something about the Shining Architecture that would translate so well into a full immersive 3D experience. Spooky as hell!!

Cageman
04-19-2010, 04:34 AM
PS it reminded me of this short by Joseph Kosinski - http://www.youtube.com/watch?v=CunBip6tcMY There is something about the Shining Architecture that would translate so well into a full immersive 3D experience. Spooky as hell!!

You are somewhat late... ;)

http://www.newtek.com/forums/showpost.php?p=1009550&postcount=52

prometheus
04-19-2010, 06:42 AM
Mr Rid..
was this calculated on a 32 or 64 bit system?

Michael

Infinite
04-19-2010, 11:12 AM
You are somewhat late... ;)

http://www.newtek.com/forums/showpost.php?p=1009550&postcount=52

It depends how you look at it... That CG shinning clip was made about 4 years ago :hey:

Mr Rid
04-19-2010, 04:06 PM
Mr Rid..
was this calculated on a 32 or 64 bit system?

Michael

64. I think I read that RealFlow 4 is still x32 in some aspect. The x64 install insisted on going in the x86 folder, and I think it doesnt utilize more than 2gb of ram, or something. I wasnt paying attention to ram usage when I was in RF.


That was incredible, the fact that the 24fps version makes it look ilke a real life scale model is amazing. ...

Newtek! a great Gallery submission here!...

I think the 24fps version is the only render that starts to look real. That and the 140fps render were taking so long to render they only use 7 pass AA which crawls a bit in places, and the overhead is only 5 pass since it was taking even longer. Lighting was setup to look decent only from the front view, and the overhead render was an afterthought where it looks too chocolatey, but it was taking too long to keep futzing with.

Mr Rid
01-03-2013, 10:00 PM
Well, somebody did it much better. I cant find what software they used.

http://www.youtube.com/watch?v=JpZI_mRC13Y

safetyman
01-04-2013, 05:45 AM
LOL -- I think that software is "Sketchpad Pro" with the "Ketchup" plugin.

ksnoad
01-09-2013, 03:27 AM
Sketchup?

shrox
01-09-2013, 08:05 PM
Well, somebody did it much better. I cant find what software they used.

http://www.youtube.com/watch?v=JpZI_mRC13Y

Looks pretty fake to me, the particle size is all wrong. Something about the background is off too...

Netvudu
01-10-2013, 03:16 AM
I think you are too picky guys. It´s a perfect replica of the original and in some aspects it clearly surpasses it. MrRid´s versión has nothing to do against this über polished version.

3D Squirrel
01-29-2013, 10:14 AM
A very interesting thread and a great scene you've recreated. From the 2013 promo's I've seen on Realflow it looks like you'll be able to boost the particle count. I like your second camera angle, it really adds to the scene. Nice work.:thumbsup:

Mr Rid
01-29-2013, 03:29 PM
... From the 2013 promo's I've seen on Realflow it looks like you'll be able to boost the particle count. ..

I see sims in the new RF using 40+ million particles on a similar system, that only took maybe 10 hours. So now, I guess I should be able to run WAY more particles in less time, and it could look much closer to the real thing.

erikals
01-29-2013, 11:51 PM
wonder, would you have chosen regular simulation in RF then, or the new Hybrido 2 ?

3D Squirrel
02-02-2013, 02:06 PM
Can you bring in an older scene to the new RF and just crank up the particles or is it more involved?

Mr Rid
02-02-2013, 02:23 PM
wonder, would you have chosen regular simulation in RF then, or the new Hybrido 2 ?

I'd have to spend considerable time experimenting...


Can you bring in an older scene to the new RF and just crank up the particles or is it more involved?


Its never that easy. For one, the behavior changes drastically with any considerable change in particle count. It will mean more glitches and weeks of messing with. If wet maps are still wonky in LW, its not worth revisiting.

Mr Rid
02-09-2013, 01:39 AM
I read that Igor Zanic did this with Naiad. No wet maps.

https://www.youtube.com/watch?v=yqriAMFXnMc

Mr Rid
03-05-2013, 04:02 PM
I happen to know someone else developing a complex fluid solver but cant say who. :angel: They want to trial it on my Shining geometry. And no, it wont support LW.
The fluid solver I was referring to was used for the Room 237 trailer

I read that Igor Zanic did this with Naiad. https://www.youtube.com/watch?v=yqriAMFXnMc

It was actually done by particle master, Rif Dagher.

Bobross
03-28-2013, 12:59 PM
Would it be possible to only upload all the scene geometry in obj (or some other universal format) so others can try some fluids sims?

Sadly I don't have the skills to model a simple hallway.

Mr Rid
03-28-2013, 04:46 PM
Would it be possible to only upload all the scene geometry in obj (or some other universal format) so others can try some fluids sims?

:goodluck:
113126

Bobross
03-28-2013, 09:07 PM
:goodluck:
113126

Dude,

You. Are. Awesome!

Thank you so much! Going to have so much fun playing with this!!!!

:bowdown:

shrox
03-28-2013, 09:48 PM
:goodluck:
113126

That is crazy cool of you.

Mr Rid
03-29-2013, 07:47 PM
WAY more particles are possible now, but I will probably never revisit this. It took many iterations to get the simple door frame adequate as there keep being leaks, or the liquid would deflect off in an odd direction, since I dont know what kind of edges were on the real thing. My set was also scaled down by half to match the miniature used in the movie, and it was pitched toward camera 2 degrees to keep the sluggish particles moving across the floor. RF wetmaps never respected LW uv correctly for some reason. Subdividing the mesh a lot was minimizing wetmap errors on another project.

Bobross
03-29-2013, 08:23 PM
Again thank you for the scene, I've been playing with it and have a couple of questions if you don't mind?

How did you get the fluid to shoot out sideways initially? Did you just fill the elevator as a volume and open the door? I'm not getting the same effect, its just dumping out straight...

I though I read somewhere that in the actual film the elevator door does not even slide open as it normally should have..

Any tips would be great! Thanks again!

Mr Rid
03-30-2013, 01:32 AM
In my scene, the collision door motion is matched to the motion of the door in the film which does not slide open all at once, and therefore influences the behavior. And yes, the particle filled elevator then pours out.

First I retimed the trailer shot by speeding it up by X6, in order to approximate 24fps. To save considerable time I ran sims at 24fps, and slowed the temp renders or previews down by X6 to view in Fusion. When the liquid behavior looked about right, I then switched to 140 fps in RF.

Later I read that the actual set was about 4 feet high which confirmed my guess that it was at half scale. And I read an article by a journalist who questioned the effects supervisor (uncredited) on Shining, who reported they did 3 different takes (not as many as is usually rumored).

I cant imagine how to engineer a real sliding door with a seal that is water tight enough to hold back over 1000 pounds of water, yet can slide open easily without leaking. My theory as to why the door in the film stutters open, is that a door edge seal tore loose as the door began to slide open, and it wedged between the door and frame momentarily. If you look closely, a dark 'strap/seal' can be seen dangling back and forth from the edge of the door. Whatever it is, it must be a mistake, along with the door not sliding open smoothly, contrary to those who maintain that Kubrick films have no 'errors' in them (others have found many continuity errors in Shining). I am sure this was difficult to shoot repeatedly, and that even Kubrick had to compromise with half-scale (he must have insisted on full scale at first), and the door not opening smoothly. I have probably studied the blood-elevator shots from the movie more than anyone, and after all the debate over how this shot was done, I seem to be the only one who has noticed the dark strap thing. But there are three different angles of the same take, used in the movie and the trailer.

I spent two months running sims to come up with fluid settings at the right pressure and viscosity for 1.6 million particles, at half scale. You'll have to figure out settings that work in your scene.

shrox
03-31-2013, 10:17 AM
I...contrary to those who maintain that Kubrick films have no 'errors' in them (others have found many continuity errors in Shining)...

There are plenty of 'errors' in 2001, like when Bowman has to get back in the Discovery through the emergency hatch. When the door on the pod blew, the pod should have flew away from the Discovery, or at least swung to the side if it was still gripping the control to open Discovery's hatch. All minor stuff though, it's the entertainment I am looking for.

Iain
03-31-2013, 01:38 PM
contrary to those who maintain that Kubrick films have no 'errors' in them (others have found many continuity errors in Shining).

Kubrick was obsessive about quality but I'm sure he himself never made such a claim.
For every 1000 generic Hollywood directors there is a Stanley Kubrick and those who spend their time finding fault in his work are really missing the point. He may have been annoying to his actors as a constant perfectionist but that doesn't mean he missed little things here and there and it in no way detracts from his success in constantly breaking new ground and producing outstanding imagery and, most importantly, classic films.

Mr Rid
03-31-2013, 04:53 PM
I was thinking more about errors that are interpreted as intentional hints at some obscure meaning. I've had such a long debate with Rob Ager and his supporters about the mysterious 'objects' he believes are hidden in the blood, who insist that every tiny detail or seeming continuity error in a Kubrick film MUST have deliberate meaning.

But even experienced critics and scholars unknowingly project their own mood or psychological bias into what they believe a filmmaker may be intending, with colorful to ridiculous speculation, as the Room 237 doc delves into.

In the original Day the Earth Stood Still, its easy to imagine the filmmakers alluded to an analogy of Klaatu to Jesus- a wise man with miraculous powers who comes to us from beyond, is persecuted, executed, and resurrected to save mankind. But in the dvd commentary, director Robert Wise insists that thought never occurred to them as they were just busy trying to get the picture done and never gave it that much thought.

I remember reading an interview with someone who worked on Eraserhead, amused over some published analysis of the film that speculated on the meaning of why Henry steps in a puddle near the beginning with one foot, then in the next scene he changes the wet sock of the other foot. But the interviewee explained that the two scenes were shot over a year apart and it was really just a matter of no one on set remembering for sure which foot it was.

Bobross
03-31-2013, 06:16 PM
In my scene, the collision door motion is matched to the motion of the door in the film which does not slide open all at once, and therefore influences the behavior. And yes, the particle filled elevator then pours out.


Hmm I'm still having trouble. Would you say the initial sideways effect is more due to fluid setting or the way the door is opened or both? I can't seem to get the particles to seep out sideways. Did you coax them out any special way or is this all 100% gravity? Did you use any daemons?

I notice your geometry is not as straight forward as I would think (box with hole and door that slides sideways) is there something going on there I should be aware of? Maybe something got screwy during the OBJ conversion?

Thanks again!

-Bob

*EDIT* I've re-watched the video several more time and the blood isn't going as sideways as I though I remembered. Still having trouble though...

Iain
04-01-2013, 03:24 AM
I was thinking more about errors that are interpreted as intentional hints at some obscure meaning.

Ah sorry, I misunderstood.
I haven't seen Room 237 (I prefer to read about these things-deconstruction and deep interpretation ruins a lot of the magic of cinema for me,) but I'm sure some of the clues alluded to are accurate. Why not? Games are peppered with hidden secrets and double meanings and its surely just as easy to write them into film (probably easier in fact.)
But ultimately, it does seem like the start on a slippery slope towards mysticism and inappropriate adulation.

Regarding the Eraserhead story, I've always thought that a lot of Lynch's weird touches were errors or just eccentricity. You can look too hard deeply into anything.

Mr Rid
04-01-2013, 03:28 AM
The liquid should pour slightly to screen right, simply due to the sliding door being recessed back further on the right of the opening, than the frame edge on the left of the opening. I only used gravity.

I checked the geometry closely before posting, and it is all positioned correctly.

djwaterman
04-01-2013, 04:43 AM
The personality type that insists on hidden meanings and coded messages in every work of genius is the type that just doesn't get it, so because they can't actually appreciate the work on the simple level that was meant they go searching for other things to justify the iconic status of the work. As if Kubrick of all people would be trying to confuse his audience, disorient them sure, but he wouldn't waste precious time and resources on things that were so obscure that they would not be noticed by general movie audiences. There is one Shining documentary out there that ends up being a front for the whole Kubrick filmed the fake moon landing idea, which is just a front for the Man didn't land on the moon anti American movement. The other documentary out there is just an examination of the maze like and deliberately deceptive design of the set, with halls and doors that lead nowhere, I'm okay with that, but that could just be the economy of set design where you build for the camera without concern for functionality. Kubrick was obsessive about detail but not stupid about it.

Bobross
04-01-2013, 04:59 PM
The liquid should pour slightly to screen right, simply due to the sliding door being recessed back further on the right of the opening, than the frame edge on the left of the opening. I only used gravity.

I checked the geometry closely before posting, and it is all positioned correctly.

Ok thanks for the insight.

I was referring to this geometry:
http://s15.postimg.org/y5rd2cakr/Untitled_1.jpg

Is that intentional? It is creating leaks for me

Mr Rid
04-02-2013, 03:36 AM
I was referring to this geometry:
http://s15.postimg.org/y5rd2cakr/Untitled_1.jpg

Is that intentional? It is creating leaks for me

When I open the obj contained in the posted zip, it does not have those gaps. All polys face inward, and all points are connected. So I dont know why it is different on your end. Did you process or convert the geometry in some way?

This is how it appears in Modeler-
113232

Bobross
04-02-2013, 12:26 PM
Hm strange. I just used default OBJ import into Maya. I'll check the settings.

ActionBob
04-02-2013, 03:13 PM
Top PROPS to Mr. Rid... Not only for sharing something that was a lot of work for him, but also sticking around and answering questions. I am sure you are not told enough that it is appreciated, especially for those of us that don't normally do fluid simulations, but get to mess around with a well built scene.

-Adrian

allabulle
04-02-2013, 03:16 PM
+1

Mr Rid
04-02-2013, 04:42 PM
Hm strange. I just used default OBJ import into Maya. I'll check the settings.

I think I see whats causing the difference.
113248

See if this works.
113247

Bobross
04-02-2013, 07:41 PM
You have "corrected them sir" :)

That did the trick!

I was thinking about doing a hybrido version but not sure if that would work.


Top PROPS to Mr. Rid... Not only for sharing something that was a lot of work for him, but also sticking around and answering questions. I am sure you are not told enough that it is appreciated, especially for those of us that don't normally do fluid simulations, but get to mess around with a well built scene.

-Adrian

+1 QFT!!

Thanks again Mr Rid!

Titus
04-02-2013, 11:32 PM
For some strange reason, I've been unaware of this thread. Amazing video! Contrary to some comments in the youtube site, I think the work is almost perfect, and the slow motion makes it believable.

Mr Rid
04-26-2013, 01:49 AM
http://welcometotwinpeaks.com/wp-content/uploads/purgatory-jared-lyon.png