Could the EX1 finally be the one for me? Almost...

[Mike Thomann]

Each year I keep finding myself amazed with the latest developments compared to the previous year's and I keep finding myself tempted to buy into the current latest developments afforded to me by not buying the previous year. After some consideration I usually settle with something that will get by, like the wonderful Canon HV20 last year. So the question I have to ask myself as a person of many interests and provider of three children, what will it take to trust that I will be content with a purchase long enough for the cost to be worth it? If the cost is $1K, I expect to be content with it for a couple of years and last year the HV20 has suited me very well in that regard. Likewise for me if the cost is $5K, I expect to be content with it for 10 years, and so on. Now the question I have to ask myself is, when will the developments have reached a point at which I could trust that I would be content indefinitely? I know the short answer should be a resounding "impossible", but I do believe that it is feasible and likely to be achieved in my lifetime as long as I don't drink too much. So what would make it feasible? The answer is when technology affords imagery and sound projected within a large living room that meets the acuity of my eyes and ears. As technology continues to evolve, there will be perceptible differences along the way. At some point, it will no longer be possible for my senses to perceive further improvement, and I will buy, buy, buy!!!!!

Acuity of my ears...
Current sound offers all of the frequency response and sampling rate I can perceive, but 16 bit has to go. After hearing the difference between 16 bit and 24 bit sound with projects I've worked on with effects that increase dynamic range, I can't believe that I ever considered CD audio as even decent. The difference is so dramatic that a 96kbps mp3 from a 24 bit original source sounds way better than an uncompressed 16 bit wav from the same source. I don't believe I could perceive any difference beyond 24 bit and I'm not sure if I could perceive a difference between 24 bit and say 20 bit if I were to able to compare. It may be the case that 24 bit makes no difference straight from a single modern mic before effects and I would be interested to find that out. As far as number of channels, I don't think that I could perceive much more than five discreet channels, but two is definitely not enough. As recording devices with 5+ channels are available and can be looped directly into the capture process, having onboard sound is not imperative.

Acuity of my eyes...

RESOLUTION:
After much reading, I've concluded that 1080p exceeds human vision acuity in the largest of home theaters at suitable viewing distance. I'm not interested in the development of increased resolutions like 2K, 4K, Ultra and so on. My eyes will never see it unless I'm closer than I expect to typically be. My eyes will see however, increased dynamic range, latitude, low light sensitivity, color depth, temporal space (Hz), and focal control. Even in a movie theater these things would play a much bigger role in visual perception than 4K over 1080p. For the same sensor size of the same technology, 4K would compromise too much of these things over 1080p, especially low light.

DYNAMIC RANGE:
Although the sources I've read are varied, the generally accepted dynamic range of human vision is 30,000:1 or 15 stops in any one scene (30 stops adaptive over 4 seconds). With many sources claiming static vision as low as 11 stops I'm thrilled with current affordable HD video with over 10 stops, but as long as we can perceive more, technology will eventually provide it.

LATITUDE:
After much reading I'm left confused about the limit of our ability to perceive latitude (something like 5 or 6 stops which we interpret as double?), but I'm led to believe that current video sensors offer plenty enough latitude before compression.

LOW LIGHT SENSITIVITY:
Let's just say this is an area that technology has a long way to go to match our vision, and hence is probably the area that I'm most interested in seeing improved. I wouldn't expect video sensitivity to match our vision but it sure makes for an incredible consideration. Some specialized sensors already have better low light sensitivity than our eyes so maybe it could happen.

COLOR DEPTH:
Ultimately, I can perceive minute tonal differences across a single hue up to 32 bit color depth. Across multiple hues I really can't notice anything beyond 24 bit. But when diffused, I surely can't even tell anything beyond 16 bit even across a full color pallet. 8 bit is definitely noticeable and I haven't tested anything in between, so I'd hope that I couldn't notice much beyond the current 10 bits available today. Also should consider that for say x bit final output, more than x bit source is required to allow room for color processing in post without degredation to x bit final output. This means that if final output x bit is to match our perception, the source should ideally be more than our perception.

TEMPORAL SPACE:
I've read that the motion picture industry intends to adopt 60p in the future as opposed to 24p. Being addicted to games I've learned that I can perceive up to 80 Hz when I really try hard, and not surprisingly I've read that that's pretty much the limit for the average Joe, with very few approaching 100 Hz. 60 Hz would pretty much be considered real-time. No more compromising combinations of motion blur and jitter to create movement.

FOCAL CONTROL:
Our eyes can do extraordinary things in an instant without our awareness. Such ease could never become available to the camera operator but I'd be happy with the ability that 2/3" sensors provide. Of course there's the adapters like Brevis etc, but I prefer to have focal ability similar to 35mm without having to adapt with more glass and less sensitivity.

Long-time industry professionals might consider these things laughable but looking back at all of the developments over even just the past few years, and considering what's happening with chip development, 35mm digital high-speed photography, more efficient encoding, and increasing storage capacity, these things don't appear all too far from becoming a reality. The wait will likely have more to do with companies holding on to each step along the way to optimize profit, but there's always competition to help things along. How much would such a camera be worth to me now? Considering that it would be my last camera, probably $30K. I know this is not possible at any price now, but that's what it would be worth to me now, so I will wait. In the meantime I will remain a fence-sitter considering a compromise with what I want and what's currently available. Last year I almost went for the Canon XH-A1, and this year the Sony EX1 is more tempting than ever. The EX1 could bridge the gap with almost all of my wants if the HD-SDI could output 1080p30 or if it had HDMI to do so. If it weren't for that, I would buy it. I wonder if Sony deliberately made 1080p30 digital output impossible to not compete with their much higher priced cameras. So I will remain a fence sitter and enjoy my HV20 and keep watching for what's to come. 1080p60 can't be too far away. Maybe the aniticipated new Red Scarlet will be unresistable. I'm aware that current display is unable to reproduce what I'm looking for but it will eventually and I'd like to begin recording uncompromised even if it is before it can be displayed.

[Eric Pascarelli]

After they come up with a camera that does all that, they can continue to make it smaller and lighter for awhile. They can then make it easier to use. And they can continue to change the recording media.

They can also make it 3D.

Then they could create something that shoots a full panorama all the time and you select your shot later. And resolution could be increased way beyond human vision so you can zoom in for extreme closeups in post.

And there are many more things that could be added beyond those to continue the process of improving cameras.

So you may as well buy the EX1 now.

[Mike Thomann]

I considered 3D but at this point I have a hard time believing I would be apt to use it nearly as much because two large format lenses is pretty impractical and most people would react oddly to Terminator 4 looking them squarely in the face.

Also considered full 360 pano and but I'm more interested in being able to tell a story through the eyes of the camera with focus.

Hmm, being uninterested in these is more reason to jump on the EX1 now. I will be happy with the HV20 for another five years if I have to, and I can't keep from thinking about what I may be able to get within that time if I don't go fot it now. And I thought considering the A1 was difficult to turn down last year. Who would have thought we would be considering something as great as the A1 before it came out? Who would have thought we would be considering something as great as the EX1 before it came out? I anticipate being surprised again soon. I don't think Sony will top this within five years unless they have to. If no one else tops it with their next line up fairly soon, I would probably less expect to be surprised within this time and would probably just get it.

[Ali Husain]

Originally Posted by Mike Thomann

After hearing the difference between 16 bit and 24 bit sound with projects I've worked on with effects that increase dynamic range, I can't believe that I ever considered CD audio as even decent. The difference is so dramatic that a 96kbps mp3 from a 24 bit original source sounds way better than an uncompressed 16 bit wav from the same source.

it would be very interesting to the codec community if you could get a 96kbps mp3 to sound better than a properly dithered 16-bit master!

Originally Posted by Mike Thomann

As far as number of channels, I don't think that I could perceive much more than five discreet channels, but two is definitely not enough. As recording devices with 5+ channels are available and can be looped directly into the capture process, having onboard sound is not imperative.

if you move your head, you can tell the difference between 5 and >5 channel systems quite easily. some research has shown you need ~1million channels to recreate a soundfield; and still there are issues. i can dig up the papers if you're interested.

a trivial test to show you a problem is: (1) pan a sound such that the amplitude is even between two speakers, (2) put your head centered so that you perceive a phantom center channel, (3) now turn your head 90 degrees, (4) the phantom image disappears. there are holes like this in any multi-channel system.

Originally Posted by Mike Thomann

DYNAMIC RANGE:
Although the sources I've read are varied, the generally accepted dynamic range of human vision is 30,000:1 or 15 stops in any one scene (30 stops adaptive over 4 seconds). With many sources claiming static vision as low as 11 stops I'm thrilled with current affordable HD video with over 10 stops, but as long as we can perceive more, technology will eventually provide it.

some sources will tell you human vision gets about 10 stops. they could be wrong though, and nobody says how they've measured this.

[Mike Thomann]

Originally Posted by Ali Husain

it would be very interesting to the codec community if you could get a 96kbps mp3 to sound better than a properly dithered 16-bit master!

I worked extensively with Sony Vegas to enhance and add effects to music and most of the time dynamic range was greatly increased for better sound. I would export the result to 16 bit wav to be mp3 endoded using CDEX with the latest and greatest LAME encoder which has been debated as the highest quality encoder even among the highest-end expensive encoders. I never thought that 24 bit would have made any difference. Then I learned that the mp3 format does not have bit depth because it's practically a vector format and so I realized that a higher bitrate source may improve the final mp3, so I exported the compliled sound from Vegas into 24 bit wav and encoded it at various mp3 bit rates to test it with the lowest being 96 kbps. I then compared each to the 16 bit wav source compiled from Vegas, and even the lowest 96 kbps mp3 encoded file from the 24 bit source sounded considerably better than the 16 bit wav source. The difference was so great that I regretted ever having worked with 16 bit prior. I didn't believe my ears so I compared the encoded file to the live output from Vegas and although it I could notice some compression, it carried the same depth that was heard in the live output. Then when I compared the 16 bit wav source to the live output, the 16 bit wav source sounded very flat in comparison.

Notes that if you feed a vector format enough source detail, the result could easily be much smaller AND better than a mapped format of some lessor bit size source. The same is true for converting pixel-mapped images to vector as well. When converting mapped images to vector, it's interesting to see visually how drastic the diminishing returns are for increased settings, especially as the accuracy of the source increases. This is due to vector efficiency. I would go as far as to suggest that mp3 be considered an improved recording format and be directly vector manipulated without any decompression/recompression stages. We only think of mp3 as compressed because we have to convert it from an inefficient mapped format and continue to unconvert/reconvert it every time we open it and save it in a mapped format world. There are now programs that edit mp3 in vector, but are unable to include mapped format without conversion. It would have been nice if all digital audio had already begun as vector format, and 16 bit mapped audio would be considered highly compressed and inefficient.

Originally Posted by Ali Husain

if you move your head, you can tell the difference between 5 and >5 channel systems quite easily. some research has shown you need ~1million channels to recreate a soundfield; and still there are issues. i can dig up the papers if you're interested.

a trivial test to show you a problem is: (1) pan a sound such that the amplitude is even between two speakers, (2) put your head centered so that you perceive a phantom center channel, (3) now turn your head 90 degrees, (4) the phantom image disappears. there are holes like this in any multi-channel system.

Thanks, I always wondered about that. I know there are 8 channel systems that can be integrated but I think for my purposes 5 discreet channels is enough. As audio is upgradable outside of the camera I'm not really going to consider it a matter in deciding what camera to buy.

[Ray Bell]

Originally Posted by Mike Thomann

Acuity of my eyes...

RESOLUTION:
After much reading, I've concluded that 1080p exceeds human vision acuity in the largest of home theaters at suitable viewing distance. I'm not interested in the development of increased resolutions like 2K, 4K, Ultra and so on. My eyes will never see it unless I'm closer than I expect to typically be. My eyes will see however, increased dynamic range, latitude, low light sensitivity, color depth, temporal space (Hz), and focal control. Even in a movie theater these things would play a much bigger role in visual perception than 4K over 1080p. For the same sensor size of the same technology, 4K would compromise too much of these things over 1080p, especially low light.

The true meaning of resolutions beyound 1080 is more than trying to stuff 4K onto a screen with 1080 pixels....

Take one of the 1080 digital TV sets and play some footage that was created on an editor with the presets of 1080 pixels. The footage is now
mapped pixel to pixel of 1080 pixel lines on the vertical axis....

Now take some 4k footage and place it onto the timeline of an editor with the same preset of 1080 pixels.... see, there are now more than 1080 pixels to map... so you are now presented with a very cool option for the timeline.
you can staticly ''crop" the 4K image down to 1080 or you can dynamicly
" zoom " and " Pan " the 4K image down to 1080 pixels....

If you want examples to see this watch some of Phil Blooms footage... in some cases he will shoot both 720 footage and 1080 footage, then he edits
on a 720 timeline and pans 1080 footage into the shoot.

So for example, you can set up a camera on a tri pod, leave the head stationary, shoot 1080 footage, bring the 1080 footage into a 720 timeline,
pan and zoom the 1080 footage (keyed) and what comes out in the final render is a perfectly panned/zoomed footage that for all purposes looks and feels like you panned the camera on the best fluid head in the world... and in actuallity you didn't move the camera at all.....

So with this in mind.... a 4K camera gives you that much more size lattitude to crop,pan and zoom further into/around the scene....

If you want you can even duplicate the exact same footage on the timeline and work with different pan/zooms/crops and you will get the illusion of a multiple camera shoot, when infact you only used one camera....

So, I'll take a 4K camera anyday over a 1080 camera.... much much more control over the final results...

[Mathieu Ghekiere]

Nobody wants a 4k camera to show a dvd to their friends...

4K is meant to be projected on the big screen, where all those pixels have a good purpose.

Of course, if you buy a 4K camera, often you are already buying a camera that has bigger dynamic range, better codec, ...
In the case of RED also 35mm DOF.

But that's also normal, if you compare the price of a EX1, with the price of RED, the cheapest 4K camera on the market.

[Mike Thomann]

Originally Posted by Ray Bell

The true meaning of resolutions beyound 1080 is more than trying to stuff 4K onto a screen with 1080 pixels....

Thanks for those great points. The fact that 4K would offer so much more creative control and variance of filming vs final output is way more significant than just being more resolution than 1080. If one were to choose between 1080 and 4K of the same sensor size and level of technology, it would take some consideration to determine whether that added control is worth the hit in the other areas of quality that the higher resolution sensor would result in. From this point on, these other areas are the primary aspects that hinder real perception, so until these are able to be fully optimized I would probably favor these over more resolution than we visually need even if it were to offer such control. Although there are so many important things to consider, I just can't resist my motivation for a perceptively real image over anything else at this point. Now if the anticipated Red Scarlet were 2/3" sensors similar to the Red, but in 1080 instead of 4K and with better low light and more dynamic range, that's my ticket.

Originally Posted by Ray Bell

If you want examples to see this watch some of Phil Blooms footage... in some cases he will shoot both 720 footage and 1080 footage, then he edits
on a 720 timeline and pans 1080 footage into the shoot.

I've watched all of Phil Bloom's shorts many times over and I try to analyze as much I can how and where he applies his techniques. I'm amazed at how quickly and naturally he does it. It's amazing how he pulls off perfectly capturing an emotion in shots that can only be done once and without a pre-run, like The Tram. I look forward to seeing where that talent goes.

Originally Posted by Mathieu Ghekiere

Of course, if you buy a 4K camera, often you are already buying a camera that has bigger dynamic range, better codec, ...
In the case of RED also 35mm DOF.

I believe I read here that the EX1 had more dynamic range than the RED. I also haven't been able to find a low light example in which the EX1 sensitivity doesn't appear better than the Red, but I've only seen a few.

[Mathieu Ghekiere]

Originally Posted by Mike Thomann

I believe I read here that the EX1 had more dynamic range than the RED. I also haven't been able to find a low light example in which the EX1 sensitivity doesn't appear better than the Red, but I've only seen a few.

I've read that too actually, and I think that's very unlikely. But, as someone that's going to buy an EX1 next week, I would sure be pleased to hear that.
I think the RAW file of RED lets you tweak it much more and get much more dynamic range if you post process it well, but of course, I don't know for sure, never worked with RED before.

[Ray Bell]

Originally Posted by Mathieu Ghekiere

4K is meant to be projected on the big screen, where all those pixels have a good purpose.

Commercial Digital Projectors (DLP) that are utilized in large screen theaters are only 2K in resolution. And even at 2K the output is manipulated(fuzzed up) so the audience will never
see the actual pixelation on the screen...