Project Requirements - Harvesting High Quality Still Images from Recorded Video

[tumtumsback]

If I am not posting this in the right place, please direct me elsewhere. I will try to keep this short and sweet.

Client wants to record wedding ceremonies, and the biggest requirement is that the solution allows for High Quality frame-by-frame recap. The idea is that an export of the ceremony can be provided to the newlyweds (USB stick, OneDrive, whatever...), and the newlyweds can later go through the footage on their own, to extract and save high quality still photos/frames. It's being explained to me that the end goal would result in the creation of physical prints (8x10, 11x17, etc.) of said photos/frames.

I am no expert, but the first things that come to mind for me are shutter speed, fps, resolution, amount of ambient light, etc. I imagine the distance between the cameras and the object of focus also matters a lot; based on the dimensions of the room, these cameras are to be mounted anywhere from 2-10 feet from the people who are getting married.

My initial plan, unless swayed otherwise, to be to use an NVR with PoE cameras. I just don't know which aspect(s) matter most -- shutter speed? FPS? Res?

Attached are images of the wedding area. I used green arrows to indicate the height at which I had planned on installing the cameras (~7 ft. from floor). It will be four cameras total, one on each corner (sorry for some reason I only took three photos of the area although there are 4 "poles" total). I believe there is usually much more light in that area, but at that time, there was a party going on (blurred in the background) and the lights were cut to this area.

Any help here would be GREATLY appreciated.

 

[wittaj]

While these cameras are good, they are not professional quality camera good. And for relatively still shots at that distance an iPhone will produce better quality.

I would not recommend these types of cameras for wedding photos.

You would probably exceed the cost of a decent professional camera to get a POE IP camera of comparable quality.

 

[tumtumsback]

While these cameras are good, they are not professional quality camera good. And for relatively still shots at that distance an iPhone will produce better quality.

I would not recommend these types of cameras for wedding photos.

You would probably exceed the cost of a decent professional camera to get a POE IP camera of comparable quality.

While I wholeheartedly agree with you, it's a bit more complicated. Without going too far into detail, one of the explicit requirements I am being given is that no photographers be present in the immediate area of those getting married.

That being said, I am wondering if there are any 'think outside the box' solutions here -- potentially something like Trace Up, or Pivo, or some new A.I. + PTZ/autofocus tech I am not aware of.

My budget here is +/- $5000 if that helps.

Another ask is that this same camera system be used for livestream capabilities, for those who cannot make the wedding.

I know a lot is being asked of me, and I realize it's quite possible that this could just be a case of "customer wants something unrealistic" ... whatever solution I provide, it need to meet the needs appropriately. The last thing I want to do is have these people spend $$$ for something that doesn't end up cutting it, so it might just mean explaining to the client that the quality of the 'stills' aren't going to be nearly as good as a professional photographer.

 

[looney2ns]

I've recorded several weddings for family members, and sometimes a "hired" videographer was there as well.
I worked with one or two cameras, One being hand held, and the 2nd strategically placed on a tripod, and could be moved as needed.
The paid guy, generally had one video camera, stood in one place most of the time.
I got up close and personal with my videos, and moved to where the action was as the circumstances called for.
Lots of ill's can be corrected during editing.

Guess who's videos everyone liked better.

Recording the way you are proposing would make it possible to miss important moments.
And end up with LOTS of dead air recordings with nothing worthwhile in them.

 

[tumtumsback]

I've recorded several weddings for family members, and sometimes a "hired" videographer was there as well.
I worked with one or two cameras, One being hand held, and the 2nd strategically placed on a tripod, and could be moved as needed.
The paid guy, generally had one video camera, stood in one place most of the time.
I got up close and personal with my videos, and moved to where the action was as the circumstances called for.
Lots of ill's can be corrected during editing.

Guess who's videos everyone liked better.

Recording the way you are proposing would make it possible to miss important moments.
And end up with LOTS of dead air recordings with nothing worthwhile in them.

For any typical wedding/situation, I agree with your approach. However, to reiterate, one of the requirements the client is giving me here is that during the ceremony, while the officiator is officiating, no photographer/videographer is allowed to be present in the immediate area of those getting married. This is one of the major drivers here. The solution has to be 'automated.'

I am being told by the client that these ceremonies are typically 20-30 min long. So, regarding your points about the important moments / dead space, I don't see that being much of an issue considering the nature of this particular style of ceremony. As for the livestream, I think I'd have a good handle on that -- I would recommend piping the RTSP stream to YouTube using something like Open Broadcast Software (OBS).

 

[wittaj]

With a budget of $5k you might find a camera capable looking at the high end lines for conferences, but an "NVR" as we use probably wouldn't be the right choice for the recording.

Remember these are surveillance cameras, not DSLR cameras. For example, you can see individual hairs and skin pores with DSLR photography equipment and you won't with these kinds of cameras. These are for a different use and different expectations.

• Sensor Size - a full frame DSLR sensor size is 864mm^2; whereas a 1/2.8" sensor popular on many cameras would be 20mm^2, so the "real" camera can collect over 40 times more light than a surveillance camera. And this doesn't even account for less light available for an 8MP versus 2MP for the same size sensor.
• Shutter Speed (Exposure) - Taking a picture with a "real" camera, you can slow the shutter down to 1/2s or longer for a nice clean picture of a person not moving.
• Aperture - With a "real" camera you focus on a specific part of the field of view, while a surveillance needs to focus on things in the foreground and background, which means the aperture is smaller, further compounding the light issue.
• Compression - A single 8MP image from a "real" camera could be upwards of 5MB of storage. In surveillance cameras, if you record at 15FPS, every second of video could be 75MB or more, which could equate to 6.5TB per day per camera. Obviously most are not going to have that kind of storage, so lossy compression algorithms are used to reduce storage and network bandwidth requirement, and that can add noise and would make blowing up these stills problematic.
• Environment - a "real" camera is used mainly under ideal conditions, whereas a surveillance camera is going 24/7 in every type of element, so the design and size impacts its capabilities.

 

[tumtumsback]

With a budget of $5k you might find a camera capable looking at the high end lines for conferences, but an "NVR" as we use probably wouldn't be the right choice for the recording.

Remember these are surveillance cameras, not DSLR cameras. For example, you can see individual hairs and skin pores with DSLR photography equipment and you won't with these kinds of cameras. These are for a different use and different expectations.

• Sensor Size - a full frame DSLR sensor size is 864mm^2; whereas a 1/2.8" sensor popular on many cameras would be 20mm^2, so the "real" camera can collect over 40 times more light than a surveillance camera. And this doesn't even account for less light available for an 8MP versus 2MP for the same size sensor.
• Shutter Speed (Exposure) - Taking a picture with a "real" camera, you can slow the shutter down to 1/2s or longer for a nice clean picture of a person not moving.
• Aperture - With a "real" camera you focus on a specific part of the field of view, while a surveillance needs to focus on things in the foreground and background, which means the aperture is smaller, further compounding the light issue.
• Compression - A single 8MP image from a "real" camera could be upwards of 5MB of storage. In surveillance cameras, if you record at 15FPS, every second of video could be 75MB or more, which could equate to 6.5TB per day per camera. Obviously most are not going to have that kind of storage, so lossy compression algorithms are used to reduce storage and network bandwidth requirement, and that can add noise and would make blowing up these stills problematic.
• Environment - a "real" camera is used mainly under ideal conditions, whereas a surveillance camera is going 24/7 in every type of element, so the design and size impacts its capabilities.

Well said. I am going to have to dumb all of this down a bunch for the client so they can truly understand that they shouldn't be trying to use this system for both livestreaming, and for producing good stills. I am going to use a spork analogy.
A spork makes an okay-ish spoon, and it makes an okay-ish fork, and although technically you can twirl pasta with a spork, it might not produce the best result.

The requirement that no photographers be present in the immediate area of those getting married is a very difficult one to work with. I realize the clients expectations are unrealistic, I just have to be delicate in my explanation.

I appreciate the help of you all!

 

[wittaj]

Yeah live streaming should be OK. Those stills of quality a married couple would expect is the hard part.

 

[tumtumsback]

Wow, okay -- so this client actually wants to proceed here, even though I explained that these 'still' photos are NOT going to look professional whatsoever.

That being said, I am trying to spec out the right Camera(s)+NVR system for the job. As the pictures attached to the first post on this thread show, and I'll also attach some more pictures to this post, the area itself is not big, maybe 10ft by 10ft. I want to ensure I can get proper angles, and capture the full body, head to toe, for those in the frame. Not only do I feel like I am going to need to get creative with where I mount these cameras, but I know there are many other important factors to consider; style (dome/turret/bullet), focal length, MP, FOV, etc.

Depending on where within this 10x10 area they are standing, they could be as close as 1-2 feet away from the camera. Am I correct in my thinking that I should focus mostly on getting cameras with the appropriate FOV and/or focal length? I have also seen many dual-lens cameras on the market too, some which claim to have different size lenses within the same camera body, and I don't understand how those work. Will each of those be their own "channel", and technically show up as two different cameras on the NVR? If so, this could be cool, as it could help to increase my chances of capturing shots with value.

As always, I am grateful for all the help I get from you all. This particular job feels far from typical, so I am trying to improvise as best as possible here.

Thanks!