The actual demo of Terravision is indeed mind blowing if you remember how powerful computers in 1994 were: http://www.joachimsauter.com/img/big693.mp4
I don't mind dramatization, but 4.5h seems quite steep for such a simple video that seems like it could be covered in a 20m Youtube video. Do you have a rough review if you ended up watching more of it if it's just really dragged on or actually contains a lot of content?
It was also a nice way of teaching my wife about what the world of computers was like in the 90's.
I found it funny - in a “how ridiculous/silly” kind of way - that at 4:40 we get to look inside a building by entering through its windows, but here we are in the 21st century with that exact capability. Individuals providing photos and/or footage of the interior of their commercial space does not seem so silly these days.
In 1994, SGI workstations could pull off this kind of stuff rather effortlessly, the only "tricky" part was to get the LOD algorithm right.
So not that mind-blowing in practice.
It would only amazing if Terravision ran on a stock PC (did it?).
I was one of the creators of Electronic Arts 3D Atlas which shipped in spring 1994. It sold two million copies on Mac and won many awards.
EA was the publisher but we developed it in London working at a BBC spinoff called MMC and getting the planetary data from UCL. This UCL unit eventually spawned Planetary Visions, a startup we worked with to supply data.
That wasn't even our first 3D zooming spinning satellite globe product. We did a huge touchscreen installation for the Kunst und Ausstellungshalle museum in Bonn, Germany. It was called "Erd Sicht" (Earth View), and in the early 90s had amazing satellite Earth graphics and animations.
Probably the biggest unsung algorithmic hero of GE is the image processing and serving pipeline. The actual client tech itself was largely worked out by the time Keyhole was bought (hence the high age of this lawsuit), and the biggest upgrades to the client came mostly in the form of optimizations and support for new features like 3D buildings. What wasn't visible was the enormous amount of work in the infrastructure required to actually process and serve the whole globe's worth of imagery. At the time, Keyhole's processing and serving pipeline just didn't scale to the amount of imagery actually available for purchase. That's no knock on their engineering skills: imagery is expensive and given their business model there simply wasn't a need to process all the imagery that existed. But once Google bought them, they were given a blank cheque and told to quite simply acquire it all. The infrastructure of the time (MapReduce, BigTable etc) were a perfect fit for this problem and allowed a rapid drain of the imagery backlog that had by then accumulated. Over time more and more of the calculations done to the raw sensor data (like ortho-rectification) were pulled in house for cost reasons and to allow further scaling - the satellite imagery providers could easily become CPU bottlenecked themselves.
This sort of article/movie seems ultimately to have some sort of ideological driver, given that its opening thesis is "for every winner, there are lots of losers; the people who blazed the trail for the others to ruthlessly make billions from their efforts". I mean, that's not true is it? This blasé statement implies the tech industry is negative-sum but it isn't even zero sum, it's positive sum: the whole reason it's successful is that it's a massive creator of wealth. The assumption that the big winners in the tech industry got there simply by aggressively stealing everything is so far from reality it surely makes the article closer to propaganda than anything else.
(I haven't seen The Billion Dollar Code or listened to the podcast yet, but apparently the character Juri is (loosely?) based on Pavel: https://twitter.com/pavel23/status/1447622978859061252 )
Poor dialogs, poor acting, lots of technical bla bla makes hardly any sense.
However, Google didn't seem to copy any of their tech but just implemented the same idea (a searchable/zoomable globe/map), which doesn't strike me as surprising as the idea isn't too farfetched and the tech is very much related to 3D graphics and game development, which had already come pretty far by 2004.
So the trailer implying they actually had it first and Google couldn't have done it without them seems pretty peculiar to me.
“Terravision was a joint project from Weathernews International (Hiro Ishibashi, Andreas Schneider), ART+COM and DeTeBerkom. The series is a fictional adaptation of the events at that time and arose from the court transcripts as well as conversations with Joachim Suater, Pavel Mayer, Axel Schmidt, Gerd Grüneis, Martin Sibernagl.”
The show left out Keyhole, which actually wrote Google Earth before Google bought them. It was written 100% from scratch and based on different and better technology.
https://avibarzeev.medium.com/was-google-earth-stolen-7d1b82...
The series portrays these guys as the heroes, but in reality they just seem to be patent trolls who fortunately lost.
A lot of really cool retro tech stuff to spot though, they certainly weren’t lazy with this even if the politics of it seem weird.
It's of course okay to express this opinion. But given all the facts on the table dismissing them as mere patent trolls seems quite unfair. I encourage you to read up a bit on these people, who are/were i. a. accomplished art professors and programmers.
Like apple came from that shit, heck the entire state of hacking right now with corporation simping don't even get me started on "bug bounties" and "Responsible" disclosure
good job.
Also the founder of Keyhole, John Hanke, came from the MUD world and now responsible for Niantic and the Pokemon game. A true visionary of blending the real and virtual worlds.
This PDF file is encoded PDF format version 1.3 witch was released in 2000. It is coherent with the informations stored in the PDF heather fields telling us this file was created 02.02.2000, 10:20 and last changed 02.02.2000, 18:20 by Acrobat Distiller 4.0 .
So at lest we can say this PDF file here is not an authentic one from 1995, witch of cause not necessarily means there could not be another PDF file somewhere else(?).
This PS still contains same dates. On footer dates January 26, 1995 while the first page says April 22, 1994.
I also found the old SRI TerraVision homepage captured at May 16, 1997: https://web.archive.org/web/19970516142933/http://www.ai.sri...
As well as the old datacenter page referring to SRI Terravision captured at June 5, 1997: https://web.archive.org/web/19970605180747/http://www-itg.lb...
And here is the newer homepage of the later program version SRI TerraVision II captured at December 02, 2000 still containing some short promo videos of the software: https://web.archive.org/web/20001202100300/http://www.tvgeo....
I need to check this site more deep but on first impression the SRI TerraVision I in 1997 and SRI TerraVision II in 2000 as shown in the linked videos are kind of inferior in sense of performance and capability compared to Art+Com TerraVision in 1994. Especially the videos from 1997(running on SGI) looked really bad and more like an early prototype. Just the 2000(on PC) videos SGI TV looked a bit better. All the TerraVisions programs may share the same name for whatever reason(?), but I would assume they do not use the same Algorithms. If Keyhole/Google "borrowed" ideas for Google Earth I would guess they did not copy them from SRI TerraVision, but from Art+Com TerraVision. But how knows?
If Brian from SGI was the connection to Keyhole/Google he could be also the connection to the MAGIC project from SRI since they use the same SGI Onyx at the beginning as Art+Com.
Why did it have the same name as the original Terra Vision?
The same name is a really good question. But since SRI TerraVision and the Art+Com TerraVision both seems to be started at the same time around 1993 and both use the same SGI hardware (an Onyx) I could image that Brian from SGI was in contact with both teams.
```
import gis
gis.renderGlobe()
```
the actual tiling logic, data storage requirements, performant rendering in something like a shitty 1990s browser... I don't think you have an appreciation for the difficulty of bootstrapping this ex nihilo.
(possibly Thomas Edison, who I believed was accused of stealing IP a lot)
Yeah, IIRC, those were the high-end of the SGI line, very beefy machines for that time, especially for I/O and graphics, so I stand by my comment.
[EDIT]: And thanks for clip mapping reference. For anyone interested in dynamic LOD algorithms, here's the paper:
"This work was supported in part by the Advanced Research Projects Agency under Contract F19628-92-C-0071."
Which implies that it was in development long before the date on the PDF.
Earth mapping was a very hot field in the 1980s with most of it being funded by intelligence agencies and classified.
That's a 6-dof spacemouse, still widely used today in CAD and 3d modeling.
What an interesting input device.
"Lau further testified that, at the SIGGRAPH '95 conference, he performed live demonstrations of SRI TerraVision to at least 500 people, and in fact “gave [ ] the source code to TerraVision” to Art+Com employees who were in attendance and “walk[ed] them through the source code.”"
1: https://caselaw.findlaw.com/us-federal-circuit/1878050.html
The court case was based on the patent date? which would be later?
I just spent 3 hours on listening to a (German) podcast on the matter with one of the people who developed the algorithm that was copied (addressing scheme for storing and adressing the different LOD tiles in memory, coincidentally the exact scheme Google uses till today, because the guys spent a lot of time optimizing).
The prior art TerraVision had the same name but was a completely different software (military simulation if I understood correctly) and didn't make use of that algorithm.
They still convinced the jury : )