Looking for photographers/storyboarders for a game

[Gregor]

A long long time ago, on a forum far away, I introduced my intent to write ZEE for the Pandora.


ZEE is Zoom-Enhance-Extrapolate, an image-based maze game parodying 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 is essentially a maze game 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. The overall goal is to find clues towards solving some crime within the images.


ZEE has been a project idea of mine for a very, very long time, and now that the Pandora has actually shipped at me, I'd kind of like to revive it (again). Here's the problem: I am a perfectly competent programmer, and I think it's a sound game idea, but I am NOT a perfectly competent photographer (nor do I have a sufficiently good camera), and I am a perfectly incompetent storyboard creator.


From past attempts what I have now learned is that it would be much, much superior to find someone who is willing to take pictures AND storyboard. If you can't do both, be wary that whatever you do will impact the other of these two roles.


Requirements for a photographer: Relatively good camera and skill using it. Ability to take pictures in raw mode (NO lossy compression (JPEG), I have to interpolate this stuff!), preferably ability to stitch panoramas or at least a sufficient understanding to take the photos and allow me to do the stitching. Enough free time to take pictures of seemingly-random nonsense, and a willingness to release those pictures for use with ZEE (or under a more open license if you'd prefer). Ideally the ability to take pictures which by framing and position seem off-the-cuff, yet are actually high-quality and have good detail and fidelity. This last requirement may be lax thanks to suspension of disbelief.


Requirements for a storyboarder: Creativity (duh). An understanding of the basic concept of the game, the requirements that places on photographers, and how to fulfill these requirements while still creating sufficiently diverse ideas to make a compelling game. Note that most of storyboarding is not creating in-game story segments, but figuring out things like what scenes should be used, what the clues within them are, and how that all ties into an overall plot. In the immediate, I'd just like to make a one-level demo, so this is a bit reduced.


You do NOT NEED a Pandora (yet) to do either of these. Technically you don't even need one to test the game (I'd be happy to provide it for you in a non-Pandora state).


If you think you have at least some ability to do both, that would be 150% awesome. If you can do one or the other, post here and maybe we can get three of us to work together (since that sort of teamwork so often works out X_X).


Here's once again to desperation,


- Gregor Richards


EDIT:


Adding some more details since people seem to be wildly-unnecessarily interested in mundane details of data storage and formatting.


Everything below (in this post) is open to change, it's just the current idea+implementation.

• The INPUT images need to be on the order of 15 or more MP (the more the better), and RAW to be interpolated well.
  
• The images as stored on the system, for use by ZEE, are JPEG-compressed, and split into 640x480 tiles. There are five zoom levels, with the resolutions 640x480, 1280x960, 2560x1920, 10240x7680 and 20480x15360. There are 341 total files per picture. The total size of a given picture should be around 15MB: they are not compressed at particularly good quality, and don't need to be (they're not interpolated, just displayed straight.
  
• The fundamental functions are Zoom, Enhance and Extrapolate.
  Zooming is just that: given an image at whatever zoom level you're at, you can click and hold to zoom in, using poor (linear or no) interpolation.
  
• Enhancing just causes it to load the appropriate tiles for the zoom level you're actually viewing. This is just an effect, it could just as easily be automatic. It's possible that the number of enhances will be limited.
  
• Extrapolation links a 3D box (X,Y,Zoom) to an action. The actions are typically either getting a clue or moving to another image. Moving to another image is, from the perspective of the game, extrapolating a reflection. Clues can be in any image linked from the original image.
  

[*]This game does not intend to actually do the magic used in CSI and the ilk, only to simulate it to whatever degree is actually possible. That degree, from the tests I've done thusfar, for feasibility sake, is slightly better than the demo I linked below (much better if somebody with a better camera helps out).


[*]Photographs are big and have flexibility issues. The alternative most often presented (using ray-traced photorealistic scenes) is completely impossible. Sorry guys, it just is. Making one photorealistic scene takes a long time, making dozens takes an eternity.


[*]Nobody involved (yet!) has particularly great photo-editing skills. The amount of photo-editing necessary is hardly stellar, but if somebody wanted to lend their abilities in this capacity they would not be refused


[*]If you have a camera, maybe you can spend some minutes while about your usual routine seeing if you can figure out a "case" one might solve with such a system that you could build out of scenes around you.

 

[Blue Protoman]

I'd be willing to help.

 

[gadgetoid]

What are the sensible requirements on camera equipment? I get some good pictures out of my Lumix FZ38, but it's no SLR.


I guess for the "zoom" aspect of the game, it would require close-up photography of the subject being zoomed, which is then interpolated between the larger scene to produce a half-step enhance effect which would otherwise be impossible.


That said, most cameras can take a picture in a high enough resolution to create a fairly good "scene" on an 800x480 screen. Even better if you use stitched panoramas.


This, of course, reminds me of that Red Dwarf episode. I could totally see how it could work to a limited extent, but how do you prevent the game becoming a "pixel hunt"? There are also only a limited number of steps you could feasibly zoom whilst maintaining suspense of disbelief.


Red Dwarf throws disbelief to the wind by starting with "uncrop".


The rotating of reflected scenes in Red Dwarf is also kinda cool. It's, by far, the best example of how this sort of thing could work- even though it's created in jest.


You'll need to throw in a lot of false positives, I guess, to truly create the concept of a "maze", this will mean photographing the living hell out of a scene.


I see why you've had so much trouble making this a reality- the content of a game is always the most difficult part to produce, and this idea in particular relies on very difficult to produce content, lots of planning and lots of ingenuity.


You're going to need story-boarders first- and I imagine the way to start this would be with written flowcharts which document the correct route(s) through a scene to find and piece together the clues. Then you add the red herrings in afterward, based largely on the photographers instincts when visiting the scene.

 

[Gregor]

What are the sensible requirements on camera equipment? I get some good pictures out of my Lumix FZ38, but it's no SLR.

Depends on a lot of other factors. My intent is to zoom the images up 300MP; in my experience, a decent 15MP raw image (about the lowerbound of what I can get as a total amateur by panorama from my dinky 7MP) which is fairly crisp w.r.t. the subject in question will interpolate well up to this, but it's extremely sensitive to being a crisp, clean, well-lit shot (y'know, all the usual suspects). I also do benefit from having a relatively small screen.

I guess for the "zoom" aspect of the game, it would require close-up photography of the subject being zoomed, which is then interpolated between the larger scene to produce a half-step enhance effect which would otherwise be impossible.


That said, most cameras can take a picture in a high enough resolution to create a fairly good "scene" on an 800x480 screen. Even better if you use stitched panoramas.

This is more where the extrapolation aspect comes into play. The game gives you 32x zoom (from the initial 640x480, mind you) before it cuts you off, but "extrapolate" is the magic potion by which you can jump from image to image (by whatever magic that is). That allows, in my opinion, a good balance between suspension of disbelief and achievability, since it doesn't have to interpolate one photo against another.


That being said, the simple tiling system I made for this is fully capable of both techniques, I'm just trying to settle on what would make a good game and be possible for photographers, while also being relatively close to the expectation w.r.t. the old cliché.

This, of course, reminds me of that Red Dwarf episode. I could totally see how it could work to a limited extent, but how do you prevent the game becoming a "pixel hunt"? There are also only a limited number of steps you could feasibly zoom whilst maintaining suspense of disbelief.

The extrapolation system I have (which is all the truly BS magic in the game) basically connects a box in 3D space (where Z is, appropriately enough, zoom) to an action, such as "find clue" or "switch to this image". Choosing the appropriate Z level to make it NOT a pixel hunt is sort of a final, image-editing task.

Red Dwarf throws disbelief to the wind by starting with "uncrop".


The rotating of reflected scenes in Red Dwarf is also kinda cool. It's, by far, the best example of how this sort of thing could work- even though it's created in jest.

(Don't know the episode). It's also far more than my initial goal

You'll need to throw in a lot of false positives, I guess, to truly create the concept of a "maze", this will mean photographing the living hell out of a scene.

Mmmm, yes and no. Subtle image editing is perfectly allowed, so it's more like photographing the living hell out of /several/ scenes, then applying the appropriate reflective properties to imply that they are the same "scene" as seen through reflections. It's again a matter of goals: The CONCEPT is the old cliché, but the GOAL is a game, it's OK to fail to live up to exactly what we see in the movies if it's still fun

I see why you've had so much trouble making this a reality- the content of a game is always the most difficult part to produce, and this idea in particular relies on very difficult to produce content, lots of planning and lots of ingenuity.

Yuh

You're going to need story-boarders first- and I imagine the way to start this would be with written flowcharts which document the correct route(s) through a scene to find and piece together the clues. Then you add the red herrings in afterward, based largely on the photographers instincts when visiting the scene.

That's a nice theory I debunked a while ago . Basically what I get is


<storyboarder acting alone> But I don't know what kind of scenes the photographer will have at his disposal.


<photographer acting alone> But I don't know what kind of scenes the storyboarder will be able to use.


I think that if I could just get the ball rolling, so we started to understand what sort of scenery was good, then it could all work, but I also think that /both/ are needed to get started.

 

[DaMummy]

i remember when talk first started on this, my only question, with the pics being such hq, 15mp as you say, how much space do you expect this game to take up?

 

[Poem58]

I have a canon 20D DSLR (8MP). I'm told I'm pretty good with it, although I am admittedly not a learned photographer in any capacity. If there's something you need feel free to shoot a PM of what you need shots of and I will see If I can help. Taking pictures has to be the simplest of the things I can do to help this community. Not sure exactly of what you are trying for but I'm assuming you are talking about a 300x Zoom not a 300MP image right? One way to do this would be to take a picture of the scene, then while still on the tripod zoom into a preselected portion of the scene and take a new image at full resolution of said smaller portion, this would allow even greater zooming range with detail, especially on such a small screen. (Example, take an image of a road with cars, need to zoom into license plates (with altered plate numbers of course, I would use my 18-55mm lens backed to widest shot, take pic, then switch to 100-300mm lens and take pic of license plates of cars in scenes. Voila! All you need to do is make the magic of transitioning between the two images for that "zoom and enhance" effect.


Again, let me know what you need. I will do what I can. Then when my Pandora arrives, I tackle learning pixel art! (There's a Dizzy out there waiting for me to gain the ninja skills I need)

 

[Gregor]

i remember when talk first started on this, my only question, with the pics being such hq, 15mp as you say, how much space do you expect this game to take up?

A lot 8-D. The way it's tiled up, ~15MB per scene. Although the INPUT must be raw, the DELIVERABLE is JPEG-compressed, which is fine so long as the ratio is OK and it's compressed at each zoom level. So probably on the order of 50MB per story section, maybe a gig or so for the whole game. It will not be a small game by any reasonable scale.


I should mention that I expect stitched photos (I don't want to keep saying "panoramas" since that's really a different thing, but that's the usual context of stitching) will probably be necessary to get the best quality, but I have a fair deal of experience stitching photos, so a photographer who understands the technique should be perfectly happy to just dump the photos on me, I'll be happy to do the stitchwork myself.

I have a canon 20D DSLR (8MP). I'm told I'm pretty good with it, although I am admittedly not a learned photographer in any capacity. If there's something you need feel free to shoot a PM of what you need shots of and I will see If I can help. Taking pictures has to be the simplest of the things I can do to help this community. Not sure exactly of what you are trying for but I'm assuming you are talking about a 300x Zoom not a 300MP image right?

I am talking about a 300MP image. My current strategy (which is perfectly adjustable, but seems reasonable for the time being). Specifically, the images are 20480x15360. However, this is AFTER interpolation. I have a small (old, featureless) demo at http://codu.org/projects/zee/test/zee.html that may be able to give you an idea of the scale. The image on that page was taken by my 7.1MP digital camera, panorama stitched up to 30MP, then interpolated up to 300MP. There is no absolute "correct" input size, and I've taken some 7.1MP images that actually look pretty decent interpolated all the way up, it's more about the quality and clarity of the input image than its size.


... though of course, bigger is better


The zoom level, for what it's worth, is only 32x relative to the original 640x480. (The remaining 160 pixels, for those wondering, are used for the on-screen menu). A 300x zoom would require 26 gigapixel image to be able to zoom anywhere

One way to do this would be to take a picture of the scene, then while still on the tripod zoom into a preselected portion of the scene and take a new image at full resolution of said smaller portion, this would allow even greater zooming range with detail, especially on such a small screen.

Imagine doing it at this zoom level (but bear in mind that zoom and focus are distinct) for a mozaic covering the entire image: This is the absolute ideal (panoramic photos stitched together). I am quite capable of stitching images like this, but both my camera and my free time are insufficient to take pictures like this. The photo on the URL above is an example.

(Example, take an image of a road with cars, need to zoom into license plates (with altered plate numbers of course, I would use my 18-55mm lens backed to widest shot, take pic, then switch to 100-300mm lens and take pic of license plates of cars in scenes. Voila! All you need to do is make the magic of transitioning between the two images for that "zoom and enhance" effect.

The problem with a "directed" technique like this is that it may be too obvious from the end result where you're supposed to go. I suspect, anyway.

 

[chosen]

http://www.youtube.com/watch?v=KUFkb0d1kbU

https://www.youtube.com/embed/QkcU0gwZUdg?feature=oembed

 

[liquidphantom]

I'll tell you now going from 15MP to 300MP image will lose you a hell of a lot of detail regardless of the enlarging process you use, there simply isn't enough information between pixels to be able to do this. Seen as the the size of the screen is small though viewing a 21mp image would give you plenty of 'zoom levels' from full screen to 100% zoom, could even go to 200% if you have good enlargement software like Genuine Fractals or Enlarger Pro although at most you could go to around 125-150MP which is about 600-800% of original size, but like I said fine detail will be lost.


That said I have an SLR capable of 21MP images and a nice selection of lenses including a Very wide angle and a Macro.


Give me a brief and I'll see what I can come up with.


To do stitched 360 deg panoramas you need a specific type of tripod mound to keep the focal plane in the same place for each shot otherwise you end up with parallax shift (distorted lines where the images join), unfortunately even the cheapest ones are bloody expensive and I don't have one.


To take several photos of one scene to be able to zoom in to it will be a considerable effort though, your looking at possibly thousands of photographs which then need to be fully edited.

 

[chosen]

https://www.youtube.com/embed/Vxq9yj2pVWk?feature=oembed

 

[WizardStan]

I'll tell you now going from 15MP to 300MP image will lose you a hell of a lot of detail regardless of the enlarging process you use, there simply isn't enough information between pixels to be able to do this.

The way I understand what he wants to do with this (and gregor please correct me if I'm mistaken) is that he starts with a high res 15MP image which is then blown up to huge 300MP proportions. Yes, it will be blurry as you zoom in, but that's kind of the point: when you zoom out to 1:20th of the image, it still looks as sharp as the original image. But then you zoom in a little: somewhere in this image is another image, a sharp image that someone has taken from that angle that is made to look like it is reflecting off something in the original image. Find that image among all the pixelation of the giant image and zoom in on that. Somewhere within THAT image is another embedded image that you have to zoom in on.


Stitched together properly, you could double up each one to go 4 or 5 images deep. To go any deeper, it'd have to blow up the first 15MP image well beyond even 300MP

 

[Gregor]

I'll tell you now going from 15MP to 300MP image will lose you a hell of a lot of detail regardless of the enlarging process you use, there simply isn't enough information between pixels to be able to do this.

You do not LOSE detail, you simply do not GAIN detail. The point of a good interpolation algorithm is not to magically add details to the image that would otherwise have been lost, it's to recognize distinct edges, corners and objects and maintain their distinctness in the enlarged image. A good input image, blown up to 300MP, will have sufficient detail to create an image that looks crisp enough that it doesn't take much suspension of disbelief to accept it.


See for example:

This is one 640x480 tile of a 300MP image blown up from a image which, IIRC, was 30MP, perhaps less (don't want to dig up the original right now to remember ). It's hardly perfect of course, but it's sufficiently crisp to see what's going on (including the monitor visible through the window), and IMHO looks less like a badly overinterpolated image (an effect mainly caused by input which has been lossily compressed such that interpolation is a much harder affair) than simply looking very closely at any digital picture. My opinion of course, feel free to disagree, but I think anyone can agree that the highest zoom level here is hardly a blurry mess. More importantly, if you put this image on a screen the size of the Pandora's, it will simply be too small to scrutinize it in such detail.


Bear in mind that 30MP is only sqrt(2) times each dimension of a 15MP image.

Seen as the the size of the screen is small though viewing a 21mp image would give you plenty of 'zoom levels' from full screen to 100% zoom, could even go to 200% if you have good enlargement software like Genuine Fractals or Enlarger Pro although at most you could go to around 125-150MP which is about 600-800% of original size, but like I said fine detail will be lost.

Again, you have to have a really poor interpolation to LOSE fine detail, you will simply not GAIN any fine detail, and it will be relatively less fine since it's blown up

That said I have an SLR capable of 21MP images and a nice selection of lenses including a Very wide angle and a Macro.


Give me a brief and I'll see what I can come up with.


To do stitched 360 deg panoramas you need a specific type of tripod mound to keep the focal plane in the same place for each shot otherwise you end up with parallax shift (distorted lines where the images join), unfortunately even the cheapest ones are bloody expensive and I don't have one.

How bad this effect is depends enormously on the scene in question too. The decision to try to use stitched panoramas is more out of necessity than preference, obviously a single shot from a single well-placed tripod is better if it's high-enough resolution.

To take several photos of one scene to be able to zoom in to it will be a considerable effort though, your looking at possibly thousands of photographs which then need to be fully edited.

You're overthinking the problem; I'm not trying to do magic nor am I trying to replicate perfectly the scenes in CSI, I'm trying to find a balance that is achievable and good enough to be fun. The picture I referenced above was originally stitched from ~20 source images. The backstory I've created (which has both hilarious pseudoscience AND evil government agencies!) explains that there's only so much "extra" information in the images, and eventually they'll still blur, but let's be fair: The player has to be aware of the fact that ultimately this is a game, and we can only do what we can do

The way I understand what he wants to do with this (and gregor please correct me if I'm mistaken) is that he starts with a high res 15MP image which is then blown up to huge 300MP proportions. Yes, it will be blurry as you zoom in, but that's kind of the point: when you zoom out to 1:20th of the image, it still looks as sharp as the original image. But then you zoom in a little: somewhere in this image is another image, a sharp image that someone has taken from that angle that is made to look like it is reflecting off something in the original image.

Wellllll, it can't just be thrown in their sharp, unfortunately it'll have to be blurred enough not to be a dead giveaway, but this is where the joy of the "extrapolate" button comes in: extrapolate means "do magic that has no sort of rational explanation", so it can produce an inexplicable sharp image, but only if it had something to extrapolate from.

Find that image among all the pixelation of the giant image and zoom in on that. Somewhere within THAT image is another embedded image that you have to zoom in on.


Stitched together properly, you could double up each one to go 4 or 5 images deep. To go any deeper, it'd have to blow up the first 15MP image well beyond even 300MP

Extrapolation is an action, the image you're currently in is pushed onto a stack and replaced with the new one. Remember, when you extrapolate a reflection in the movies, it no longer looks like a reflection at some warped angle, it looks clear and perfect. That's the magic of bullshit! It's also the reason why this whole concept is feasible

 

[WizardStan]

Extrapolation is an action, the image you're currently in is pushed onto a stack and replaced with the new one. Remember, when you extrapolate a reflection in the movies, it no longer looks like a reflection at some warped angle, it looks clear and perfect. That's the magic of bullshit! It's also the reason why this whole concept is feasible

So your game plan is to take one image, allow the player to zoom in, looking for reflections that might be useful, and then zoom/enhance on that reflection which effectively loads another image to be scanned for? That is much easier! I thought you wanted to stitch everything together into one single mega 300MP image: wouldn't have been the way I would have gone, but didn't want to discourage the creative process here. If you're doing it the way I understand you are now, the zoom and enhance can be done real time so you'd only have to store the original images, not the 300MP image I was worried you were planning on storing.


The problem is that a 15MP image, even with compression, to keep enough detail to make it useful, is still going to be 5 or 6MB. A dozen images and you're at 60-70MB. Depending on how complex you want these scenes to be, you could get some really big game files, hundreds of MB possibly.

 

[Gregor]

Depending on how complex you want these scenes to be, you could get some really big game files, hundreds of MB possibly.

Yup!


The size is an issue I choose to more-or-less ignore, people will just need big SD cards ... alternatively I'm considering making it load some things dynamically. I'm doing it by making 640x480 tiles, so the fully zoomed-in image is 32x32 640x480 tiles, so dynamic loading from the webertubes should be very feasible. Still, if you've got a big card, use it!

 

[lulzfish]

The reflections will add detail that wasn't in the image, you'll have to be careful that isn't obvious. I assume you'll do something like composite each reflection onto the next larger image, so the reflected scene is already there, but blurry? Then use gamma-correct blending to fade in the full-resolution reflection as the user zooms in. And the reflection probably won't be rectangular, if it's anything like that door. In fact, that door is only reflective on the windows, so you might end up with an alpha mask in there, too. And the reflection doesn't point anywhere useful, just at the ground.


I would do this with raytraced images if I were myself, and if I was doing this. Would be more flexible than photography, and then you might even be able to generate new images on the Pandora. (Or rasterizing + multipass reflections)

 

[Gregor]

The reflections will add detail that wasn't in the image, you'll have to be careful that isn't obvious. I assume you'll do something like composite each reflection onto the next larger image, so the reflected scene is already there, but blurry? Then use gamma-correct blending to fade in the full-resolution reflection as the user zooms in. And the reflection probably won't be rectangular, if it's anything like that door. In fact, that door is only reflective on the windows, so you might end up with an alpha mask in there, too. And the reflection doesn't point anywhere useful, just at the ground.


I would do this with raytraced images if I were myself, and if I was doing this. Would be more flexible than photography, and then you might even be able to generate new images on the Pandora. (Or rasterizing + multipass reflections)

The ability to figure out what's going on 3-dimensionally in a scene is ... pretty restricted for most people, it's probably reasonable to fudge the accuracy a bit for the sake of making things come together properly. And yes, this is the least well-defined part of the current plan.


As per ray-tracing ... making one photorealistic (or good enough) scene is not easy. Making dozens is borderline impossible. Now let's imagine we want HUMANS in them ... yikes. Photos are less flexible, but a hell of a lot more feasible.


PS: To all: I added a lot more details of what's currently done or planned in the first post.

 

[myownlittlworld]

If you need a photo editor or stitcher, I've had a bit of practice. I'd be happy to edit the images up if it'd help. I've done more, but those are owned by other folks and not mine to display online.

 

[gruso]

I remember ZEE, glad you're back on it.

 

[Gregor]

If you need a photo editor or stitcher, I've had a bit of practice. I'd be happy to edit the images up if it'd help. I've done more, but those are owned by other folks and not mine to display online.

Ooooh, very nice. I'll keep you in mind if I actually manage to get things going in the near-immediate.

I remember ZEE, glad you're back on it.

I was never off it ... sort of? I got an offer to write some cases from the first time I posted this, and even got some starts, but it fizzled out. It's a very slow-going project with a lot of halts ... I suppose I need to do a whole demo myself, with my stupid stories and my poor camera, and hopefully make it JUST interesting enough to attract other people. Oh well, I keep on tryin'

 

[Gregor]

Photographers in potentia: I'm looking for somebody with or with access to a house they would be willing to photograph (outside + two interior rooms and maybe a hall) and also move a couple things in before taking said pictures. The particular design doesn't matter too much, but it should be big enough for a small family, and being bigger than that would be fine.


EDIT: And I should mention, if you fit these requirements, please message me asking for further details, thankee