Game Idea, Need Photographers

[GregorR]

I have a game idea that's been rattling 'round in my brain for years (I've even partially implemented it), and I think the Pandora would be a good platform for it (touchscreen, lots of space, etc).

The name of the game is ZEE, short for Zoom-Enhance-Extrapolate. It's a parody of those ridiculous scenes in spy movies where they take some photo, zoom in on it, "enhance" it by some magic, extrapolate other images from reflections, etc. It would be a maze game, essentially, except that the maze space is a series of images: You zoom in on the current image to find a reflection you think may help you, extrapolate that to get another image, etc.

It's actually not that complicated to develop, but I constantly run into this simple issue: I simply don't have the resources or creativity to take and edit all the photos. They need to be very high quality (although not obscenely high quality; if they're taken raw and never compressed then interpolation actually makes a decent image when increased to the sizes necessary), and they need to be of fairly-arbitrary scenes (cityscapes, houses, people, vehicles, whatever).

So here's what I'd like:

1. People who think this is a good idea: some "maze" ideas.
2. People who think this is a good idea and have a good camera: Take a bunch of photos (in raw mode; this really is important) and license them in some convenient way (preferably creative commons, or at least a license to unlimited use within this given game). I can and will do the processing if necessary.
3. Anybody: Comments on what could make this game more interesting or fun, or ideas for what the overarching plot might be (if such a thing is necessary for this)

(If you're curious, the game itself would probably be written in C or maybe C++ with SDL and licensed under some open source license, and of course a precompiled version for Pandora when that becomes possible. I have no interest in selling this.)

 

[Prometheus]

[*]Anybody: Comments on what could make this game more interesting or fun, or ideas for what the overarching plot might be (if such a thing is necessary for this)

I'm afraid I can't help with anything here, but I will say this: I think with your mentions of spy movies, and reflections that may help you, and suchlike, you may have an answer. It sounds like an ideal fit for some sort of detective-related game.

 

[gruso]

Ha, awesome game concept! Zoom Enhance is an IRL meme at my house.

I don't have much input to offer right now, but I do have a question. Do you plan on storyboarding the game first, in order to give photographers certain things to capture? Or will you give let the cameras run free, then design levels around the pictures you get?

 

[GregorR]

Do you plan on storyboarding the game first, in order to give photographers certain things to capture? Or will you give let the cameras run free, then design levels around the pictures you get?

Having a storyboard and making particular requests would be great if there's an imbalance with many photographers and few designers, but I feel that it's an unrealistic strategy in the far-more-realistic scenario of "oh *whew* I managed to find two photographers willing to help me out"

 

[atomicthumbs]

I have a Digital Rebel XSi, an 18-55 mm lens, and a 50mm prime. Tell me if and what you want me to take photos of. I can put them under the BSD license.

 

[Bosbeetle]

I really like the concept but as mentioned it should have a narrative. Somehow zooming in on pictures feels to me like the old adventures using real life acting and photographs (the dame was loaded, gabriel knight 2, phantasmagoria etc.) these games might be of some inspiration.

Another idea is having a criminal sending you photographs (you know criminals alwas send clues otherwise they would just be silly) however there might be some clues obscured

But by our impecable skills we find out what was obscured and that helps you advance in the story.

 

[may88]

I have very little programming skills (scripting skills yes coding no). I am as good a photographer as I am photogenic.

But I love the idea. Sound like this one could take a lot of disk space so we all might be needing those 32Gb SDHC.

Sounds like a scene from one of my favourite films, Bladeruner. Need to avoid a SF genera for copyright reasons.
The cool bit was how Deckard was able look around corners. You zoom into a room and rotate. I guess you could have a strategically placed mirror to provide the reflection for this. How about having photos at two(or more) angles and generating a stereo view that can be rotated to find hidden clues. Advanced stuff and would add complexity to the coding and add to the photos needed.

I think you should storyboard a chain of five or so pictures and ask for samples of these and draft a test program.
Find the puppy:-> Picture 1: living room (tv sofa lamp window wall mirror) Hook location: beveled edge of mirror ->
#2: Open door leading to a hallway (more living room from other angle) Hook=metallic handle of door
#3: view down a hall with an open door at the end and a car in the distance, kids bike on the lawn. Hook=handle bars->#4a; car->#4b
#4a: false lead dead end picture of LHS of house
#4b: puppy on the lawn.

Is this the sort of idea albeit with a more exciting goal.

Need some one with Photoshop skills to comment on size/resolution the images will need to be for this. Guess we're zooming to 800x480 right
Will difference in exposure levels make it look bitty or will add charm? Can this be normalized to within a standard range?
I assume we'd zip(lossless) the raw images for storage and unzip them on the fly.

Longevity of this game is sure going to need a lot of photos. You could make it a never ending story and release in self contained episodes. Each could be a thread of the main story and can be developed independently and released when ready.

 

[lulzfish]

I've converted some random photographs into 800 by 480 JPEGs, and they averaged 130 KiB each.
So if we assume that something bad happens and each image in this game takes exactly 200 KiB, you can fit 4000 such images into the first 800 MiB of a 1 GB SD card.

However, these are just 800 by 480... you won't be able to zoom in and "enhance" them nicely, at least not in arbitrary increments.

As for zip compression, gzip and zip both resulted in a 5.3 MiB zipped file for a 5.7 MiB directory. It's probably not worth it to use zip.

 

[dflemstr]

I've converted some random photographs into 800 by 480 JPEGs, and they averaged 130 KiB each.
So if we assume that something bad happens and each image in this game takes exactly 200 KiB, you can fit 4000 such images into the first 800 MiB of a 1 GB SD card.

However, these are just 800 by 480... you won't be able to zoom in and "enhance" them nicely, at least not in arbitrary increments.

As for zip compression, gzip and zip both resulted in a 5.3 MiB zipped file for a 5.7 MiB directory. It's probably not worth it to use zip.

Use PNG; it's got a higher compression ratio for 24-bit images if you use eg optipng. I tried creating some random 2048x2048 semi-organic images and they were around 1MB each (making it possible to store 800 images using the above storage model). And it's lossless!

Also, the zooming can be done using mipmaps in vital places (kinda like Google Maps does it; load more detailed images when you zoom in, but only if they are available). Not a problem I think.

 

[GregorR]

Zooming will all use mipmaps. Which is to say, it will work as such:

There are N zoom levels, probably 5, for a maximum zoom of 2^5 = 32. Each zoom level is split into the appropriate number of 640x480 tiles for ease of loading (I think it's wise to stick with 4:3 for the sake of not having to crop the original input photos, the other 120 pixels can be used for the menu). When you zoom in, the client just keeps interpolating the image at the zoom level you started from (so it looks like crap) until you click 'enhance', at which point it generates an image from the zoom level below and above your current one, and interpolates to get a nice image. I'm writing a quick web demo (which doesn't use full mipmapping), I'll have it up with some example photos once it's at least complete enough to see how the game ought to work.

A single set of photo tiles will take on the order of 17MB. This is not a small game ... luckily we're living in the future, where SD cards can be huge

- Gregor Richards

 

[GregorR]

P.S. I like the idea of releasing episodes one-at-a-time ... probably more humanly-achievable too

 

[danielliu88888]

this is awesome. may i suggest some sort of haunted mansion mystery, or something?

 

[foxblock]

Zooming will all use mipmaps. Which is to say, it will work as such:

There are N zoom levels, probably 5, for a maximum zoom of 2^5 = 32. Each zoom level is split into the appropriate number of 640x480 tiles for ease of loading (I think it's wise to stick with 4:3 for the sake of not having to crop the original input photos, the other 120 pixels can be used for the menu).

I think it's a great idea for a game, very unique. I am a photographer myself (though currently too busy for any serious photography), so I am looking forward to this one and hope it will be as good as it sounds in concept.
Just one thing, most professional cameras create photos in the ratio 3:2 (for example my Pentax has 6M pixels, so the photos are around 3000x2000), lower budget (non DSLR cameras) tend to use 4:3, but it's actually pretty uncommon in serious photography.
So depending on how many different people will participate in this, you probably will have to crop anyway..

Still, best wishes for your project

foxblock out

 

[GregorR]

3:2??? Grr. Well, the game will probably have to have a consistent aspect ratio just for sanity sake, and 4:3 crops of 3:2 are a bit less damaging than 5:3 crops of either.

A super-basic web demo of zooming and enhancing is at http://codu.org/projects/zee/test/

 

[Japa]

I'm willing to help with photography.

I can do 3264x2448, ie 4x3

however, the raws that I manage to squeeze from my camera need some fixing.

 

[GregorR]

however, the raws that I manage to squeeze from my camera need some fixing.

Raws always need some fixing, they're still better than lossy-compressed jpegs though

 

[GregorR]

Did some camera experiments today to see how realistic this is with images I don't get by searching google for "super high-resolution royalty free creative commons pictures" . Some thoughts:

1) Images interpolate up acceptably well to about 10x pixels (about 3x in each dimension). Here "acceptably well" isn't "acceptably well for the movies", it's "things don't start looking smudgy and ridiculous, although fine details are probably lost". Wonderfully enough, anybody can take a 30MP picture if they have a decent 6-or-more MP camera with at least a 3x zoom (I barely fit these specs myself), through the magic of stitching. If people want to contribute pictures but don't want to deal with the not-insignificant hassle of stitching, I'd be happy to do that part myself, they just have to do the fancy camera-work (zoom in, keep the focus/exposure settings identical or as close as possible through each shot, make sure they have plenty of overlap).
2) Interpolation is actually pretty nice: An image interpolated 42x (about 6x in each dimension) is actually tolerable so long as the input was crisp and taken at a good low ISO value so there's not too much noise. My 7.1MP camera makes not-half-bad 300MP pictures so long as I put it on a tripod, manually set the ISO as low as is at all reasonable in the lighting conditions, and capture in raw mode. Again, it's clear that fine details are lost, but it looks believable enough.
3) The biggest problem will be believably faking reflections. I am not a GIMP/PhotoShop expert. If somebody is, speak up?
4) Choosing scenes may not be hugely difficult once a good story board is in place, but finding acceptable facsimiles of those scenes may be quite a bit harder. Hopefully those intrepid photographers who have said they might help are pretty intrepid

 

[Japa]

*downloads hugin*

right then, time for some practice shots.

btw, are cars that should have gone out of production more than half a century ago bad subjects?

 

[GregorR]

There are no bad subjects if I can find somebody sufficiently clever to do storylines for me. Ideally I'd like to do nothing other than code (maybe music), as I have no other relevant skills

 

[moxie]

I could probably contribute images, too, especially given some sort of idea of what to focus (a-ha-ha) on. I have a Pentax D-slr that'll do 3000 by 2000-ish raws.