Pull focus in post production...

[Thomas Demirian]

I was wondering. Wouldn't be possible to update the red firmware with a magic code so that you can pull focus in post production if you shoot 3d?

The reason Im wondering is because when you shoot 3d you use 2 cameras and somewhere along the line the firmware / hardware must calculate the distance between objects to get the 3d look.

Im no programmer by all means but it feels like the people here at RED should have the knowledge to rewrite the code a bit so that it stores that "DISTANCE BETWEEN OBJECTS" information. Hence make it possible to pull focus in post just like if you had a Z-depth layer.

Maybe in the future

 

[David (dudi) Namir]

:001_smile:Great idea we are sometimes rehearsing in the editing room and writing the script at the screening:thumbup1:

 

[Martin Weiss]

You need to change your screen name to your real name. See the Recon sticky for details.

 

[Eric Lange]

:001_smile:Great idea we are sometimes rehearsing in the editing room and writing the script at the screening:thumbup1:

Ohhh come on you don’t have to be suuuuuper snarky!

I have had patent examiners be far more creative and naive with certain prior art combinations (believe me).

OK Solidghost it IS possible to do what you describe but has little or nothing to do with the firmware in a camera.


So left and right cameras capture a sequence of left and right pairs of images. Each left and right image pair constitutes a stereogram or stereo-pair. If the cameras are well aligned, good optics etc. etc. the left and right images look pretty similar, except that corresponding features (on left and right images) and points may be translated horizontally with respect to each other. This is not a global translation of the right image with respect to the left image, but rather left and right corresponding subtle sub-regions of the images will be horizontally translated with respect to each other; this is one form of “parallax”. It’s these variations in “parallax” that help create the apparent variations in depth when stereo viewed. The brain and visual system is able to effectively decode the left and right images with their associated parallaxes into a fairly coherent spatial “percept”. Very broadly speaking we re-present to the eyes what they would naturally see, and the brain does the rest (there is a lot of debate and complexity related to this idea).


So for about 100 years or more the field of photogrammetry has been rather good at extracting very precise true 3d measurements from stereograms. The process generally hinges on having a well defined system, in terms of focal length/principal distance, as well as determining position and orientation of cameras, and precise ways of determining pairs of corresponding features and recording precise pairs of image coordinates relative to their respective images coordinate system. In the digital age this process has been taken a few steps further.


By various methods it is possible (using separate software that work on the captured imagery) to extract the 3d geometry recorded in a stereogram. There are different approaches to this. The software we built has one of the best manual editing stereo interfaces so that very clean and precise VR/CG models can be extracted. Auto assist is used to automatically extract points, surfaces and features using various machine vision/pattern matching techniques. However standard machine vision methods tend to fail as they are not very good at making various inferences about certain surfaces for which stereo data is essentially missing, (shadows, and weird lighting and surface effects or basic lack of texture). In this sense the human faculty for stereopsis and 3d image interpretation is far more adept. We developed a technique that can embed a stereogram into a 3d substrate that is in the basic shape of the objects but precisely and cleanly shaped to the key features. This enables the capture and presentation of extremely high complexity surfaces and visual/optical phenomena in a spatialized 3d/VR environment. These new types of models look absolutely life like (not computery at all) but it allows you to move the stereo virtual cameras to different positions and view angles. The models are conservatively 400 times more efficient at being rendered than conventional CG models that aim to represent the same level of complexity. A lot of real time applications such as video games, but visual quality is more than sufficient for very high end VFX.


So to answer your question, YES you can dolly in and pull focus (in a separate rendered environment created by the data extracted from the original stereogram). However, if the data is not in the original stereogram then essentially you can’t pull focus on something that is a blur to begin with (this is what (dudi is going on about). So you could shoot with greater depth of field and then render as shallow depth of field for the particular position of the camera. There are limits in terms of occluded features, and this is where you get into using multiple stereo cameras to complete the 3d picture, (I am however suspicious that Cameron buys 50 Epics…)

The other advantage of this type of data set (we call a Coherently Stereo Textured Model or CSTM) is that it creates a new division and type of parallaxes where the broad range of macroscopic parallaxes (for the whole scene) can be controlled and reined in to safe limits (for comfortable viewing) WITHOUT flattening the fine 3d features and 3d surface texture. (I.e. you can get very rich and engaging stereo that is comfortable on the eye and head, perhaps useful for key sequences in a movie). So essentially you can shoot with cameras further apart to capture fine 3d detail and then real time re-sample and rein in the “headache” inducing large range of parallaxes. In essence what you imagine (Solidghost, in your mind’s eye) IS certainly possible NOW, (and has been for some time); it’s just not in RED firmware yet (or right now or any time soon; and is probably not a processes that needs to be on board a camera (if you get what I mean?)).


A new area we are working on is a new type of what we would describe as a S-3d “Engine”, that takes a completely new and VERY deep approach to machine vision for 3d from 2d stereographic data sources. This is a very all encompassing method that makes more conventional techniques seem like a blind man stumbling around a dark alley with a stick by comparison. The technique is VERY compute intensive but is closing the gap between machine and human capability. Maybe 25 years from now (high res/cinematic res) cameras will have this capability on board real time, when sufficient bandwidth and processing power is available. Maybe you could imagine more clearly than I why such an on board capability would be a “must have” feature.


Some of the best inventions come from almost naive questions, and if you have an open mind you can say OK well why not? ;(then tear your hair out for ten or fifteen years to make it happen!!!!).


Hope this helps,

Eric

 

[David (dudi) Namir]

Eric, not laughing at you, laughing at the way sometimes we work nowadays and I am in the business since 1985. It wasn't like that back then.

 

[Eric Lange]

Eric, not laughing at you, laughing at the way sometimes we work nowadays and I am in the business since 1985. It wasn't like that back then.

I’ve been a kid on set since the 1970’s so I think I kinda know what you mean but could you elaborate a little (I’m not sure I entirely catch your drift)???

Are you talking about the freakish level of technology that is being brought to bear these days as compared to something that might have been solved with a well placed piece of "gaffer's" tape (in years gone by?), or the fact that people need to shoot stuff right with fewer cameras (on set) in the first place? rather than this notion (and joke) that everything can be solved in “post”.

[In the “old days” so called “process shots” were to be avoided at all cost and if required might have been regarded as some sort of failure. Nowadays everything “mopped up” in “post” seems to have very strange boundaries that are creeping totally into the production phase (and even earlier), so the joke about the violation of time and space, as it can all be "done in post", is appreciated…].

Cheers,

Eric

 

[Eric Lange]

With the “reality capture” technologies that we are developing/have developed that these look WAY better than laser scanning and are able to capture the complex surface phenomena as naturally lit. Given that things modeled and rendered in CG have a limit to their level of realism the CSTMs are spell bindingly life like and yet the workflow is very practical.

So rather ironically we are hoping that our technologies will encourage various Optical effects shops to build more physical models and sets to be recorded this way to capture all the complexity of naturally made and built physical models and objects, rather than build only in CG.

Even my kids prefer the original Star wars movies because the “models” and explosions and stuff look real and comment that the CG ones don’t look right. So In essence “we” would like to see more of a return to more traditional film craft and provide something that bridges the gap between CG and physically built props, models and sets/locations.

Essentially best of old and best of new.

Just a thought…


...But yes I agree there is much to poke fun at about how things have come to be done, but at the same time we have to take it seriously too.

 

[Fabricio Morato]

I was wondering. Wouldn't be possible to update the red firmware with a magic code so that you can pull focus in post production if you shoot 3d?

The reason Im wondering is because when you shoot 3d you use 2 cameras and somewhere along the line the firmware / hardware must calculate the distance between objects to get the 3d look.

Im no programmer by all means but it feels like the people here at RED should have the knowledge to rewrite the code a bit so that it stores that "DISTANCE BETWEEN OBJECTS" information. Hence make it possible to pull focus in post just like if you had a Z-depth layer.

Maybe in the future

Hi,

i don't think its that easy to do a "rewrite" on the software to do this... what you can get out of a 3d cameras setup is the disparity differences of the 2 which can be converted to a kind of zdepth but its not that good (even in post the results are not close to pixel perfect)...
And even if it could be done, you will need to get an image that everything is in focus, which is kind of hard with S35 sized sensor and using fstops and lenses that normal DPs use on movies....
For refocusing in post my best bet is using Plenoptic cameras: http://en.wikipedia.org/wiki/Plenoptic_camera
Adobe demoed this technology last year and looks promising...

 

[Eki Halkka]

The software we built has one of the best manual editing stereo interfaces so that very clean and precise VR/CG models can be extracted. Auto assist is used to automatically extract points, surfaces and features using various machine vision/pattern matching techniques.

This sounds very, very interesting... is your software available for purchase, is there any more info somewhere?

Also, can your software create the 3D meshes from i.e. a moving single camera shot of a static scene? One would assume this is also possible in theory, if the shot is tracked in 3D and the camera parameters are known...

 

[Eric Lange]

This sounds very, very interesting... is your software available for purchase, is there any more info somewhere?

Also, can your software create the 3D meshes from i.e. a moving single camera shot of a static scene? One would assume this is also possible in theory, if the shot is tracked in 3D and the camera parameters are known...

Yup absolutely.

Thanks for asking; it’s not my intention to push software per se, but to clue people in on what’s actually possible (or is being achieved right now). This is how we are collectively able to move forward as we are all in an ecosystem of sorts even if folks are developing things behind closed doors it’s still good to have an idea of what everybody is doing. Even SONY for example can’t operate in a vacuum. I think the forward thinking and openness of Jim/RED is exemplary. I think for smaller companies wanting to do big things they (red) serves (at least for me) as a much more progressive model than perhaps some other companies/business models. It's been pretty amazing to see and learn how all these things unfold, good and bad; Jim seems to me to be one of the most honest business men you are ever likely to come accross.


[The software and other goodies] are not quite out the door (yet). At the moment we are primarily building extremely precise cameras to act as the eyes for the software. So the first camera systems that we are building are based around 40 MP and 56 MP HDR sensors. So these are in essence medium format surveying instruments that have very precise control (it’s not precision for precision’s sake but rather when stereo is working over a precise threshold then a much cleaner and faster work flow is possible). The reason is that we seek to glue together multiple 3d fragments and pieces from different stereo camera positions (to do this right to a high standard of finish requires precise equipment of a reasonable metrological grade). This would enable us to do extremely precise full 360 degree capture of a live subject (for example) or larger scenes. The software that we have spent years developing is called “Objective Reality™” can work with dynamic stereo data sources and stills based stereographic sources. I think one of the key advantages is that we can record things from the size of a bug to a mountain with a single instrument, a single laser scanner can’t do that, one of the other basic advantages is that our techniques capture the subject as naturally lit (as on set) with real embedded 3d surface depth and optical complexity, the models are stereo textured so that they don't look extra flat artificial or out of place when viewed in stereo.


Probably later this year we will be knocking on the relevant doors when we have a complete “demo reel” and fully functioning hardware to really knock people’s socks off and leave them in do doubt whatsoever that this is very good set of tools for many problems concerned with CG and stereo cinematics etc. and other applications besides. The software is fairly mature and well tested. Along with that we will be looking to partner with organizations that seem like a “good fit”. We have (recently) granted patents on key components of the technology. Our software has a very rich array of tools but we are also looking at technology transfer and licensing. The software does work with less precise stereograms and I have also been interested in the type of imagery generated by various “beam splitter” type rigs. It seems that these rigs are getting more precise and therefore more useful to these types of processes, and the EPIC has very great promise indeed.


I guess to answer your question, the CSTM type models are backwards compatible with conventional CG wire mesh/polygonal texture mapped models. Feel free to PM me if you like; when we are ready to pull the dust sheet off (hopefully before the end of 2011) I will certainly post shots and some links to just say “see what I made.. (finally…)”. Some of the camera designs are quite different from what most people have seen before and the raw capability of the models is pretty jaw dropping.


I think when Jim Jannard and Red team manage to get around to producing a 645 based sensor solution, then this will be an absolutely killer combination with the technologies that we have developed. High res for stereo based VFX and advanced processes is really crucial (I believe).


Cheers,

Eric

 

[paulherrin]

eric. that sounds incredibly fascinating. i have heard of such technology being developed and tested... im wondering what market you guys are aiming for? is this something you call in as a special team that's invested in the technology, or can only large studios finance this? or do you believe it could be something for small studios or even individuals?

basically, sounds like something i might want to get acquainted with...

 

[Patrick OMara]

you can extract 3d data from a 2d video and use the zdepth for a lense blur/rack focus effect in adobe after effects.

 

[Eric Lange]

eric. that sounds incredibly fascinating. i have heard of such technology being developed and tested... im wondering what market you guys are aiming for? is this something you call in as a special team that's invested in the technology, or can only large studios finance this? or do you believe it could be something for small studios or even individuals?

basically, sounds like something i might want to get acquainted with...

Paul that’s a very good question.

[If I remember rightly a lot of your posts are VERY tech savvy, and if I’m not mistaken several years ago you were posting about CUDA and other such things… or maybe someone else (I don’t know)].

As you know with specialized systems (especially in the “movie” industry) there can be this situation where a company has to send a whole crew with a truck load of several million dollars worth of equipment and there are only a hand full of such systems in the world, or the software technology is part of a screamingly expensive DI suite. On the other hand you also have software that is developed for example by (let’s say) Adobe that consumers (on mass) can buy for $600.00 or $1200.0 or whatever.


The software we have engineered has been knocked flat and rebuilt from scratch about three times over seven years to produce a VERY solid platform that is also highly modular and yet has accessible tools for developers and end users alike. So we are trying to steer the middle ground between specialized behemoth and software you might acquire from a “box” shifting merchant. Our aim is to make the software (and some hardware) accessible to smaller entities, departments and forward thinking individuals. For the higher end hardware we are aiming to set up in the US and Europe some rental and service capability.


We have a few pre-orders right now for folks wanting to record world heritage sites given that the system can handle extremely life like and complex data sets very effortlessly that are also metrologically accurate. For these guys there is a very good business model, as they have a full system for a fraction of the price of Laser systems and software. The advantage is the data sets are visually far superior to laser scanning and for complex sites the time in the field and in the studio/lab are dramatically reduced. Competitively they can charge ½ as much for their services and yet have a substantially larger profit margin than their Laser counterparts. The system works with laser scanned data as well, so there’s no real conflict (i.e. it can augment existing laser systems).


In terms of markets, as you can appreciate [Paul] there are more application areas than you can shake a stick at, but as the technology is new especially the CSTMs we have to also create an new market for ourselves as well. So we are getting into government/high energy physics type applications, heritage recording etc. and obviously video games and VFX/cinematics. Each industry has a negative connotation of the other from a design and marketing point of view. So if for example we say, this would be fabulous for video games development, then the cinematic guys might think, low res and cheesy graphics. If we say this is a cinematic solution, then people assume bad metrology, overly expensive, specialized and slightly flaky systems, if we say this is very high end government/surveying solution, clunky hardware and bad standard of finish/dynamic range and image quality would be assumed, and if we say we have an Adobe plug in (licensed) or otherwise then people assume cheap “disposable” software for the masses, and “stunted” capability. The reality is that the quality of the results far exceeds all of these requirements. The hardware looks something like a cross between an anti aircraft gun and a piece of equipment sent to Mars do you spray/anodize it black so that it looks more “cinematic” and do you spray it white for the government/surveying market; similarly with the software there is a look that says DI suite…(I’m sure you know what I mean). ). Obviously Pre-Viz would be a “slam dunk”, but it’s going to take time to get into areas that make sense.


We/I have a long term view we are happy to let people (at the appropriate point) to spend substantial time with the software to learn its capabilities. So Media Tech colleges and Universities are good places for young and up and coming talent to get their head around advanced systems (with very substantial academic breaks), also and we are trying to see how we can intelligently down grade the system and break off various functional components so that most organizations can afford some sort of CSTM type capability (without causing injury to ourselves). As a lot of this is very new we have to find our way as we go and then re-assess make course corrections and so on. So again RED is a very good example of this approach.


The basic technology does solve some key long standing problems with s-3d and various psycho-visual issues and computational problems. At the moment people post “Woe is S-3d type articles” and that all of this S-3d stuff is just going to go south. I’ve been working with S-3d type data sets and media for over 20 years, and have been very aware of the fundamental problems. In the “industry” they are just starting to come up against real limitations and barriers, so we are pretty hopeful that some of the approaches we have will unlock a much greater capability for s-3d that has people saying at the box office, “that was worth it!”. i.e. a worthwhile S-3d experience.


This year we are putting together our official portfolio and “Demo Reel” so we are interested in specific projects at all levels, as we like to see that the technology is useful and is solving real problems of one sort or another. If someone has a good idea and it’s a good application then we are always interested. The software runs on two monitors both in stereo so there aren’t a lot of people out there that can even install and run the software so we have to be pretty pro-active and most folks are probably going to have to buy hardware to run the system, so it’s gonna take time.


Paul I’d be interested in your perspective on some of the things you might be interested in doing, but from previous posts I think (if I remember rightly) that you have a good handle on software and computer/computer industry issues.


Cheers

Eric

 

[paulherrin]

wow, awesome eric. thanks for sharing. so when can i beta test?

haha, i'll send you a PM, my interests are mostly for cinematic/vfx purposes... i'd like to get a better idea of how the system operates and what it could do, and what range of hardware is necessary. do you have any webpages or documents you could send me that has an overview of the features (or projected ones)? or maybe a peak at a test or two?

 

[Eric Lange]

wow, awesome eric. thanks for sharing. so when can i beta test?

haha, i'll send you a PM, my interests are mostly for cinematic/vfx purposes... i'd like to get a better idea of how the system operates and what it could do, and what range of hardware is necessary. do you have any webpages or documents you could send me that has an overview of the features (or projected ones)? or maybe a peak at a test or two?

Thanks Paul,

Probably before the end of the year (will make it so).

I/we have learnt from past experience that we have to be super careful who we show this stuff too, it’s more of a case of appropriate timing. We had one organization several years ago we let “in” to see what we were doing. At that point it was our first attempt at a production system from a fully featured prototype. The imagery etc. was/is very impressive and the “clients” were completely blown away, but even though in a lab everything works and looks great, they couldn’t understand why we wouldn’t give them the software until we had re-engineered a very solid platform from inside to out. We didn’t want to put a lot of time into making a system totally bug/bullet and monkey proof when we realized that we wanted to re-architect the software into real platform.


…They would go to conferences in spite of signing NDAs basically proclaim that what we have deep sixed other similar technologies, and yet they become totally incensed that we would not let them have a version of the software that they would run into the ground. You really felt that in another age they would have had a battering ram at the gate of our offices to storm the castle for the software. It’s a very odd thing that when people and especially larger organizations get to see something that somehow the other party has acquired some sort of notional “ownership” even though no deals exist and no money has changed hands. We’ve been very careful and have waited for our IP to actually be granted. So we act like a (little) submarine that pops up to the surface from time to time until we are completely ready to launch (I think Jim has commented that he regretted some things that he was a bit too open about in the past).


The folks that we have shown that are closer in, have pretty much been instantly sold, but that is showing imagery that is relevant to their sector. I have to admit I really hate showing work in progress, before we have collected and authored the kind of demo reel that would appeal to other areas. That’s our principal task this year apart from some hardware engineering. [In the past we used monster photogrammetric large plate film cameras]. In the film and Movie business especially, people have soooo little time, that unless your first impression is anything less than leaving them absolutely “gob smacked” then you are wasting everybody’s time. You can’t say “no no we have some really great demo data sets this time…that you might like more, more chicks, less nukes, wait…come back…”. First impression is everything, no second chances.


We used to have a website but we took it down as our proximity to Intel and Sandia National Laboratories (in New Mexico) (and style) made us seem too much like an atom smashing government spin off. The other thing is that we don’t want to give away too much for free to our competitors before we get ours out the door and in the air. This will be our first set of products to launch so we will have a fully formed web presence on that day. However I am a great believer in design, and corporate image and it’s a bit of challenge to develop a coherent style that fits, Cinematics/VFX, Government/defense and Video games, Heritage, etc. etc. as each appropriate style and mode is counter to the other’s respective disciplines and each demand a high degree of focus. I.e. Film people will eye video games development with suspicion, both view government and surveying with different suspicions and so on and then there’s broadcast. In terms of style and corporate image etc. at some point (time willing) we will try to converge on perhaps a newer styled aspect of credible media technology. I think the Germans have this worked out quite well with various media technology institutes like the HFG and ZKM and the like that have also (historically) done well in “Hollywood”.


This basic thread is an excellent illustration of the ‘mix”. So you have Solidghost has the notion that maybe these stereo rigged cameras have firmware that is computing where the objects are and maybe we pull focus in post. We have (Dudi) poking fun at how we do things now as compared to what was normal 20 years ago. Fabricio posts about a plenoptic system (that’s not super precise) BUT underscores the fact that standard approaches to z extraction in conventional post related software has been weak (in the past (this doesn’t have to be the case..)). Eki wants to know is the paradigm of meshes and polygons utilized and has twigged to the fact that well tracked camera data will fly. Others might want a Chroma key style array of XY and Z pixels. So you really get the idea that cinematography is going virtual whether it likes it or not (perhaps)…?


From all of this It’s pretty clear that the normal barriers to high production value are eroding year on year and becoming more accessible and that all of the media tech is converging so the lines between video game, pre-viz, TV broadcast, cinematic production, VFX and high end “gov” applications are all blurring. Jim Cameron has been talking about trying to have video game releases that are more cinematic so there is the possibility that he could “creatively” get to more projects than he would otherwise. Lucas for example has a whole s-3d live action Star Wars™ TV series with scripts collecting dust on the shelf BUT says that the thing is too expensive right now to make using conventional methods and that technically something else has to be devised to create a new way doing S-3d based movies.

See http://www.huffingtonpost.com/2011/05/31/star-wars-tv-show-george-lucas_n_869116.html

One of things that underpins a lot of this are the copious amounts of processing power that can be brought to bear ever more cheaply that allows various technologists to explore things that ten years ago would be absolutely unthinkable. I still get the sense that media tech is a bit of dirty word in some circles and the tyranny of the computer has skewed too much the “art” of film making… In the mean time we just get on with it and worry about “image” later on perhaps.


Cheers,

Eric

 

[paulherrin]

totally understandable eric, thanks! sounds like you guys have a lot of considerations. have you thought about doing just separate interfaces for different industries (and perhaps hardware, if necessary) and then you could at least compartmentalize the marketing a little bit... i dunno. i'll send you a pm pretty soon and maybe we can talk a little bit, i would definitely be interested in the future of this system and just getting a better idea of how it can be applied.

 

[Eric Lange]

totally understandable eric, thanks! sounds like you guys have a lot of considerations. have you thought about doing just separate interfaces for different industries (and perhaps hardware, if necessary) and then you could at least compartmentalize the marketing a little bit... i dunno. i'll send you a pm pretty soon and maybe we can talk a little bit, i would definitely be interested in the future of this system and just getting a better idea of how it can be applied.

Thanks for the suggestions.

We have done something like that: the software is modular but the principal division is between static and dynamics. So it is designed to assemble 3d fragments from multiple stereo camera positions that are synched up, synched stereo frame by stereo frame in parallel. People who want to survey historic buildings and world Heritage sites don’t really need this dynamic capability and type of management.

It’s not our plan to become “box shifting merchants” but if we scale and grow its pretty clear that we will have to set up separate divisions or even companies or a “group”. The thing that we (VERAXYZ, (pronounced veraxis) are best at is R&D, so already folks that are more service oriented are looking to get involved. It’s a bit of a chicken and egg that we really have to get our feet more than wet in several application areas to grow and find solid application areas that really serve our clients needs to the fullest possible extent. The film/movie business can be very demanding in that service is everything and the full staff capability has to be available at the drop of a hat. That is a big commitment, and we are not based in LA etc... Cinematics compared to industry in some respects are a small market so one can throw away the better part of a million dollars (or more) on development for serving a not very bankable niche market (that may evaporate by the time you get there), so we are trying to be clever about how further capabilities are developed and in particular in partnership.


Targeted, specialized web presence is important as you say. The best laid plans of mice and men; I am reminded of the famous top ten application areas that Intel had for their first processors/CPU. Home computer was not even on the list, after defense and space applications. So there are some good home markets that will enable us to grow but in my bones I feel that we also have core application areas that we have yet to discover. This is why it’s so important (I believe) to be in contact with folks from many different related and even seemingly unrelated disciplines. You never know. This has never been done before.


What we have developed is NOT a “me too” type of product and is a new paradigm, so it’s not super straight forward to come up with the usual apocryphal 1% market share statistics of this and that. There is a technological gap of sorts but I have to say I am super happy that many recent developments in Stereo cinematics are really helping to bridge that gap (in particular a lot of people and organizations that hang out here even/especially), so we don’t feel like we have to pay for the whole shooting match to bridge this type of gap for successful technology transfer. And at least in other application areas that we can/are providing complete “turn-key systems and imagers”.

So feel free to PM me and I can fill you in on the details etc, today I’m being a bad Dad on “father’s day” so better get back to it…

Cheers,

Eric

 

[IAN SUN]

Amazing stuff Eric. We would love to acquaint ourselves with these capabilities as we develop our indie S3D features.
I want to immerse my 10yld budding film making son in these emergent technologies, and image creation paradigms as we move toward our more fully realized S3D-dreamscape future.

 

[Eric Lange]

Amazing stuff Eric. We would love to acquaint ourselves with these capabilities as we are develop our indie S3D features.
I want to immerse my 10yld budding film making son in these emergent technologies, and image creation paradigms as we move toward our more fully realized S3D-dreamscape future.

Cool Dad.

My father pretty much banned me and my brother from pursuing carriers in the film industry.


I have to say the software has I think one of the best cg oriented manual editing interfaces on the planet. This is where the quality of the models really comes as for right now fully automated processes for 3d extraction are still partially doomed to failure. With the CSTM paradigm very few polygons are required as they act as macroscopic (invisible substrates to correctly dispose the 3d and s-3d surface texture correctly and precisely in 3d space.) Key edges and certain features are extracted and created to sub pixel accuracy; the models are very clean and fast to produce yet support an amazing level of life like complexity.


Unreservedly the software in terms of its principal tasks are “child’s play” to use. Normally with CG based platforms many months and years of experience are required to be useful for example with Softimage, Maya, 3d Studiomax etc. etc. That is lot of training time and such experienced modelers/animators should be well paid. I have to say that with our software a very low skill level is assumed and for the most part for the main work flow only a few days training is required providing the operator can see reasonably well in S-3d and has reasonable knack for finding surfaces. So far in testing we haven’t found any one that couldn’t do it yet and these were people that had no prior 3d/cg training or experience before. So I am not advocating a sweat shop mentality but you could hire a small cohort of college students and under supervision you really could get through a huge volume of models that are frankly in terms of quality are way ahead of the pack. To achieve similar results using conventional cg animation workflow would be far more expensive and time consuming.


A lot of time and thought has been put into good design in the interface as it relates to workflow. So an early pre-cursor of the software was designed so that conservators, restorers and archaeologists could drive it with ease; so if a UK archaeologist can drive it most certainly your ten year old can. Maybe this is what I should be doing, putting the software into the hands of ten year olds; :”look Ma, look what what I can do”, (perhaps a good marketing angle; look what this bright ten year old could do in less than two hours of training...)


It’s not my intention to make this the Eric/Objective Reality show. But I would say that in general that most technical and physical problems are solved by going to a higher dimension. So traditionally 2d composites have been all pervasive, and now over recent years the power of full 3d composites are being realized. I think the Foundary/Nuke is an excellent example of this understanding of what an explicit 3rd dimension can really do for you. The problem with s-3d (live action) when you insert CG elements that the s-3d gives the viewer the ability to analyze surfaces in a far more complete way. Usually the flatness of standard texture mapping on various models is completely betrayed and laid bare. At least with the techniques that we have, one could instead create models that in s-3d would stand up perfectly to scrutiny and are indistinguishable from other 3d/2d composited elements. This is because the models support correctly and coherently mapped stereo textures so that the surfaces of CG models have real optical depth and complexity rather than stick out like a bit of a sore thumb (when viewed stereoscopically). This has real advantages for S-3d video games that normally would look MUCH worse in S-3d.

So IAN as you say a “S3d dreamscape” future, almost by necessity to solve the problems; Cinematics are year on year going to become increasingly virtual I don’t see anything (from a physics) standpoint that will reverse that trend. I have to admit I am a big fan of the value of S-3d perhaps in broadcast (or special events digitally projected) to put you “tele-presently” somewhere else. The haggard example of Zahi Hawass opening a tomb in Egypt, the viewer has the possibility to “experience” the site in tremendous detail, basically at a distance of 7000 miles, for a lot of these sites, access is forbidden. I am sincerely hoping that some of these types of technologies can be of some cultural value or educational value to “the young”. I think there is lot more work that has to be done to reach visio-spatial types of kids that are very gifted and bright but get left behind in the more normal educational systems. I think it’s very worthwhile to have kids become spatially and visually literate and let them play and do things with these types of tools, as 15 or twenty years into the future they are going to be using systems in “anger” that we can’t even imagine right now. The types of things that these kids will be doing and using probably do not even exist today. Right now in conventional education there is very little to challenge that part of a kid’s mind.

Cheers,

Eric

 

[Eki Halkka]

I've been waiting for something like this to emerge - the time seems to be ripe, with multiple close-but-not-diamond projects out there, like Microsoft's photosynth and 3D tracking apps creating point clouds, but not solid volumes - and in the other hand, the available image modeling software needing manual labor, and laser scanners creating meshes, but without a connection to i.e. video footage, and people using game controllers to extract Z -data... with unusably low resolution and accuracy... etc.

And there you are, with a concept that not only seems to incorporate the best of all those worlds, but goes much further - so yeah, you bet i'm interested ;-)

...sent a PM too.