I have no idea what I'm doing...

[Jon Lee]

Hallo all, my name is Jon, and I've got a problem. Working on some graduate research for a university, I have been tasked with setting up a system to adequately record our experiments (possibly to win nobel prizes, but most likely not). However, of all the people to be chosen, I have the least amount of experience with recording stuff (at least video). However, I was (and am still) willing to learn as much as I can do get this right so I don't end up blowing a grand on something that doesn't work.

With that in mind, I ask the brilliant minds of DVinfo.net to help a poor graduate student out.

Act 1: The environment
• The set up for this experiment and recording is a light controlled tent (think a total blackout tent to prevent external light from coming in, and light from inside getting out). We will be recording little robots dancing around on the floor (hopefully semi-autonomously).
• The area/space I have been assigned to record is roughly 9 feet by 6 feet.
• Lighting wise, there is no overhead lighting, and no side lighting. Imagine a closet. With no lights. But it's a tent. That's what I've got. However, there will be a DLP projector projecting a pattern down onto the floor (which is kept a pristine white color). So there is light, it's just not extremely bright (or high quality, at least I don't think so), and it will be changing in color and intensity through the experiment.

Act 2: The players
• I would like to able to control the DV camera from a remote point (think outside this tent), preferably a computer. This is because the quality of the light inside the tent must be sound (for the robot sake), and it would be a pain to stand on a ladder peering through a little hole to watch the experiment through the DV camera viewfinder/LCD. So, I guess that means I would like to be able to preview what the camera sees through my computer monitor, and to be able to record/stop from the computer.
• To top that off (I know, I just keep on going), I have a fairly well equipped PC (a desktop), probably mid-range (but it is getting old). 3ghz Dual Core, 4gb ram, no external video card. Come to think of it, I've got a pretty average computer for mid 2009!

Act 3: The plot action
Questions and concerns:
• With such low-light conditions, I would like to have some semblance of manual exposure control. Is this possible?
• A corollary to that questions, are there cameras with "better" low-light image quality? That is probably a stupid questions, but I've already come out and said I'm pretty dumb when it comes to this stuff (ask me about consensus estimation theory though...).
• Will a FOV of 9' by 6' be possible with a standard lens and about 8 feet of space? I would think so, but I haven't used a DV camera like this... since 10th grade (or so).

Act 4: Fin
To recap:
• I would like something with manual exposure control, is this possible?
• Would prefer something with better low-light recording quality (the more I think of that, the dumber it sounds), possible?
• I would like to record an area of 9' by 6', with 8' of height to adjust, possible?
• I would like remote control and preview, preferably to computer, is this possible?
• Finally, I recognize I will probably need some sort of firewire card, so any suggestions? I'd like something on the lower end of the price range.

Quality wise, I'm not expecting a feature length film, this is mostly for the documentation process, as well as to aid/assist in monitoring our ongoing experiments. I don't need HD quality stuff, nor will I appreciate the hefty hefty hefty price tag.
Price wise, for all of this, camera and firewire card, and possibly additional software and stuff, I'm hoping to be under 500, if 400, that would be awesome! Those prices I kinda threw out there looking at low-mid level consumer cameras new. Of course, buying used is an option.

Finally, any camera suggestions would be great. I don't need a camera that would get me through the oil fields of Kuwait and back, but something that could do the above things (or at least most of them), stand a drop here or there. I'll be in a lab setting, but there won't be any explosions (save Skynet being sentient), toxic caustic chemicals, just a bunch of clumsy grad students.


With the above in mind, I ask you, o great minds of DVinfo, to help me. I've got a lot of questions, but like I said, I'm willing to learn. Point me in whatever direction, and bring on the heat. Thanks!

[Dave Blackhurst]

Hi Jon -
The lighting is going to be a problem PERIOD... at least as you describe it. I'm going to rule out using a camera with IR/night vision, as I presume that would contaminate the experiment...

A "remote control head" can be had for around $100 on eBay, you can probably find a Wide Angle or more likely fisheye lens so you can get the desired field of view in that tight a space. It'll depend on the camera what you will need...

Maybe there are some consumer level cams that perform well enough in low light (I seem to recall the newe Sanyo line being fairly good?), but it's going to be REALLY tough to find something that will work - I'd recommend the XR500V from Sony, which is quite good in tough lighting conditions and I don't think any other camera right now would match it, but right there you're 2X over budget... AND it won't have a live firewire feed, as it records to HDD/Memory card - I think you can get a live feed off the HDMI though, and you could monitor via any TV with HDMI - not 100% sure about this, perhaps if the composite has a live feed that would work too.

I believe that your budget and your goals are going to be in irreconcilable conflict...

[Peter Mykusz]

The only inexpensive solution would be some sort of security camera. I've seen web and security cameras that are able to be controlled with your mouse. You would record directly to your disc. I'm sure that someone in that field could make recommendations as I don't know enough about these cameras.

[Greg Paulson]

I've got an idea of what the layout is that you are describing. The robot is supposed to navigate and react to the changing patterns on the floor, like a maze, or moving objects.

If the 'walls' of this maze are black, and the area the robot crawls around on are 'white' then you'll probably have a fair amount of light for the scene.

I think your best spot for the camera would be mounted next to the LCD projecte, pointing down at the floor, since this is very much a 2D environment, looking down would give you the most accurate data.

Since the video is an important part of this project, then demand some consideration for camera placement. Some scaffolding set up on either side of the tent, then a catwalk across them, holding the LCD and the camera and yourself.

If the goal is to see how accurate the robot is, you'll need to be zooming in and keeping tight on the thing as it moves around, so you've pretty much got to be there with the camera to point it around.

And you can record right there, or firewire to your computer and record the whole thing from start to end. Probably the best idea, because you can review things (well, not you, you will be over the tent, but someone else) very quickly.

This system works the best, because you can use ANY camera , and can have a few on hand to test which works best (everyone's got a camera). And you are not relying on some specialized control interface. Trust me, you'll get screwed because the quality of image and quality of remote controls will never match. If the camera isn't supposed to move, it will need to be HD in order to give enough detail. In any case, shoot with a shutter of 15fps if light is going to be a problem.

You might look at using a DSLR with a video output, those tend to have very large imagers and will work better in low-light.

Don't use the computer for anything except recording firewire. Use a small B&W or colour TV up top and watch with that. I know the scope of the project looks huge, but honestly, it's really quite simple, even for a novice.

[Jon Lee]

Originally Posted by Dave Blackhurst

Hi Jon -
The lighting is going to be a problem PERIOD... at least as you describe it. I'm going to rule out using a camera with IR/night vision, as I presume that would contaminate the experiment...

Hey Dave, thanks for the first response! I was worried no one wanted to help me. While the lighting is not ideal, I think it is enough to get by. We have been using a system of overhanging webcameras, which are used for a CV system, but they do a fairly good job of picking up the robots as well as what they're doing. So you are correct, I definitely do not need IR/night vision, but I'm not sure if lighting will be too serious of a problem. I was mostly just unsure if certain camera's require ideal/perfect lighting. I expect most do not, but I'd like to be steered/pointed away from any that do.

I guess a better idea of my set up would be this:
Watching a projector inside an office board room, and recording the presentation screen. The room would not be lit, but there is a significant amount of backlighting from the projected image, which should light everything pretty well. If this is confusing, I can take a picture of the set up we have.

A "remote control head" can be had for around $100 on eBay, you can probably find a Wide Angle or more likely fisheye lens so you can get the desired field of view in that tight a space. It'll depend on the camera what you will need...

I looked this up on Ebay, and I don't think I will need to control the direction of the camera. I was hoping just to capture the entire floor from a fixed position, and mostly control the camera's record/stop function externally. Being able to see from an external source would be nice too (a live feed I guess). But I don't think a requirement is moving the camera around.

Maybe there are some consumer level cams that perform well enough in low light (I seem to recall the newe Sanyo line being fairly good?), but it's going to be REALLY tough to find something that will work - I'd recommend the XR500V from Sony, which is quite good in tough lighting conditions and I don't think any other camera right now would match it, but right there you're 2X over budget... AND it won't have a live firewire feed, as it records to HDD/Memory card - I think you can get a live feed off the HDMI though, and you could monitor via any TV with HDMI - not 100% sure about this, perhaps if the composite has a live feed that would work too.

Thanks for the recommendation Dave. I think the XR500V is a little too pricey, as this unit will be used only for this one project, not a department/lab shared camera, but it looks really nice. Recording in HD isn't a must, but it seems pretty much standard-fare for a lot of DV cameras, but again, it is kinda pricey. I haven't done too much research other than flipping through Canon/JVC/Sony's websites, so any further direction would be great!

I believe that your budget and your goals are going to be in irreconcilable conflict...

Hey I hope not! Gotta stay optimistic. Thanks again for your help though.

Originally Posted by Peter Mykusz

The only inexpensive solution would be some sort of security camera. I've seen web and security cameras that are able to be controlled with your mouse. You would record directly to your disc. I'm sure that someone in that field could make recommendations as I don't know enough about these cameras.

Peter, thanks for the suggestion. I think at one point we did consider security cameras, but they got tossed out because they seemed like their resolution would be kinda low. However, I should do more research, because there have to be some pretty high resolution units, I just hope they are within in range that I'm qualified to spend. I'll get back to you guys about what I think about these field. I am definitely open to any expertise in this field though, as well as as any more suggestions.

Originally Posted by Greg Paulson

I've got an idea of what the layout is that you are describing. The robot is supposed to navigate and react to the changing patterns on the floor, like a maze, or moving objects.

Pretty close Greg! We're doing research on developing consensus estimation for these robots to recreate the image we're projecting down. It'd be used for large scale sensing (such as outdoors, waterborne, etc) for chemical/temperature, etc sensing.

If the 'walls' of this maze are black, and the area the robot crawls around on are 'white' then you'll probably have a fair amount of light for the scene.

I think your best spot for the camera would be mounted next to the LCD projecte, pointing down at the floor, since this is very much a 2D environment, looking down would give you the most accurate data.

Accurate assessment, but I do not need to record data for the experiment with this DV camera. We already have a CV system in place to record accuracy and act as robot GPS of sorts, so this is mostly for documentation for further research, presentations, nobel prizes, etc.

Since the video is an important part of this project, then demand some consideration for camera placement. Some scaffolding set up on either side of the tent, then a catwalk across them, holding the LCD and the camera and yourself.

This will not be possible, we do not have enough vertical space. Likewise, the space isn't large enough where we'd have to move around to capture the entire arena in the FOV of the camera (at least I don't think so, depends on the camera FOV I suppose). Hence my questions about camera FOV. I think with 10 or 9 feet of space, to capture a 9x9 area is not too much to ask, but I have no idea about this stuff.

If the goal is to see how accurate the robot is, you'll need to be zooming in and keeping tight on the thing as it moves around, so you've pretty much got to be there with the camera to point it around.

See above.

And you can record right there, or firewire to your computer and record the whole thing from start to end. Probably the best idea, because you can review things (well, not you, you will be over the tent, but someone else) very quickly.

This system works the best, because you can use ANY camera , and can have a few on hand to test which works best (everyone's got a camera). And you are not relying on some specialized control interface. Trust me, you'll get screwed because the quality of image and quality of remote controls will never match. If the camera isn't supposed to move, it will need to be HD in order to give enough detail. In any case, shoot with a shutter of 15fps if light is going to be a problem.

Hmm... I had not considered that HD would be necessary. The robots will be fairly small, but not microscopic (yet). Right now they are about the 3" x 3" looking head down. Is HD really necessary to capture these?

You might look at using a DSLR with a video output, those tend to have very large imagers and will work better in low-light.

Don't use the computer for anything except recording firewire. Use a small B&W or colour TV up top and watch with that.

Also something else I did not consider. I am kind of limited with the number of computers, so it might be necessary that I record DV to an intermediate source (tape, card, doesn't matter, I can capture it later), but I still want to be able to preview the image without looking directly at the camera. Can I do a live feed image to another source (a spare LCD monitor, tv, etc)? That does make it harder to start/stop the recording if I can't control it from a computer though. Any other ideas for set up?

I know the scope of the project looks huge, but honestly, it's really quite simple, even for a novice.

Thanks for the support Greg! I'll keep y'all updated once I get all this straightened away the set up done. Maybe you guys can be a "thanks to" in a Nobel Prize speech (do that even do those?).

Looking forward to hearing more. This has helped me think about somethings, I appreciate all of your suggestions, etc.

[Bryan Cantwell]

Jon,

Just a tip on pricing this stuff out. Don't use the company website as a guide (sony, etc), go to B&H Photo Video | Digital Cameras, Camcorders. The list prices are much higher than the street prices!!

[Bryan Daugherty]

Bryan, you beat me to the punch.

Jon, the B&H website is good for pricing and comparing features but the best asset in my opinion is the staff there. Take some time to give them a call and talk to the sales staff about your needs. In my experience, they are very knowledgeable and will help you narrow the field to equipment that will work well for you. If possible call them when you are in front of your computer and they will help you pull up the suggestions in RT which will allow you to bookmark them (set up a wishlist) and compare features. In that price range you are looking at a narrow field. Best wishes!