4:4:4 10bit single CMOS HD project

[Jason Rodriguez]

Sorry Obin,

I'm just trying to figure out the pros and cons of each system before I think about spending the money to purchase one.

Ben's system looks very promising:

Pros:
USB 2.0 (easy to hook up to a computer)
Seems quick and easy to setup (good software)

Cons:
Color Saturation is less than ideal
Frame-rate's not locked
8-bit (I'm noticing banding artifacts)

Your system:

Pros:
10-bit
Much better color
Faster frame-rates

Cons:
Streaking?? (I really thought this wasn't suppose to happen with CMOS)
XCAP software isn't so good
Frame-rates locked?
Strobing (rolling shutter?) on slow-motion shots?

There's a lot of question marks on the cons with your system Obin because you mention stuff and problems here and there, but inquieries are either never answered/followed up, or they get lost in the shuffle of all the ideas that get thrown around on this thread.

Again, I'm sorry to be asking so many questions, but I'm trying to get to the bottom of what is good/bad about these systems. This thread is now 28pages, and I'm finding the search tools on this site less that ideal for finding the info that I want (it just spits out the thread, yippee, now I have to sift through 28 pages). I've gone through the pages just to see where things get messed up, and it seems like you start to mention the streaking problems after you started doing white balance with the gains on the camera.

This streaking worries me, because frankly CMOS is NOT suppose to do that. Streaking was a very early CCD problem, and like I mentioned from the Micron site, even they advertise no streaking. Your early pictures have no streaking. I guess I'll just have to call Steve and see what is/is not the problem.

For instance, I found after about 30 mintues of searching that you had set the gains at 11db for red and 4db for blue and green. Maybe puching up the red so high wasn't such a good idea? 11db, if that's anything like the camera gains I'm used to is A LOT of gain, especially when i try to run at 0 to -3db. 12db gain on my former DVX100 made that camera look very crappy. So maybe that high gain settings to color balance the image has thrown things off?

[Ben Syverson]

Hey guys,

I just emailed back & forth with Sumix re: CMI/Viewbits. Sumix is the engineering and design house for CMI, which sells the cameras. I assume they sell mostly to scientific/industrial concerns. However, Sumix is also interested in working with filmmakers directly to figure out what we want.

Farhad at Sumix also dropped a bit of a bombshell when he told me that they were working on improved software and firmware for the camera that would let it do 12bit log->10 bit transfer, and then some kind of conversion to map it back to 12bits, with controls over each channel. In my opinion, even 9 or 10 bit would improve our color situation, so this could be just the ticket. That could move the SMX-150c from an experimental tool to a production tool. They'll be testing it in a few weeks, so I'll keep you updated.

Around the same time they'll have a Linux driver, which I'm eager to test out. Hopefully we can get better performance from a bare bones Linux install.

Farhad and everyone at Sumix have been extraordinarily helpful and curious as to what I'm doing with their camera. They've been making optics and scientific cameras for a few years, but decided to get more heavily into cameras recently, so now they have 8 electronics and software engineers working just on cameras, with additional physics and metal specialists. They'll also be adding tech support engineers to help people with their various applications. So for them to be experimenting with the filmmaking market is kind of like two worlds colliding. But they're very excited and interested.

When I asked them about Altasens, they said they were on track for a 3650-based camera coming out in the late fall. The interface of choice? Firewire800. They tend to like simple solutions like Firewire and USB because they can power the camera as well as transfer the data. They may also release a Gigabit Ethernet version later.

I'm really stoked about the Altasens model, but I'm glad I'm working with them now. They want to know our desires and interests now, so they can try to accomodate us. The Altasens is going to be great no matter what, but we can help shape their control software and so forth if we get involved and tell them what our needs are.

Anyway, Obin was posting about lamps again, so I shot this. Hopefully this demonstrates there's no streaking on the IBIS-5.

Also, I shot a couple of test shots outside today again. I think I'm getting a better grasp on the situation. Here's a color corrected image from that test.

- ben

[Jason Rodriguez]

Farhad at Sumix also dropped a bit of a bombshell when he told me that they were working on improved software and firmware for the camera that would let it do 12bit log->10 bit transfer, and then some kind of conversion to map it back to 12bits

Whoa.

Okay, one less con for Ben's system :-)

BTW, you may have heard him wrong, he's probably talking about 12-bit linear to 10-bit log back to 12-bit linear. That's a normal workflow since 10-bit log was designed to contain the data in a 12-bit linear file, but at a reduced size-it's a nice form of compression :-)

[Ben Syverson]

Jason, you're probably right about the 12-10-12 issue, but that's how Farhad pitched it to me. Either way, as long as we're seeing at least 10bits transferred, I think we're doing okay. :)

To follow up on the frame rate issue: the camera software sets its frame rate in a settings dialog. It says something to the effect of "take a frame every ___ ms." The problem is that 1/24sec == 41.6666... seconds. So you either do 41 or 42ms. I'm not sure why it varies, but hopefully Linux will make the fluctuations a little smaller.

Also, the variations are really never more than about .3 frames a second, so the drift is:
23.6 * 90 seconds == 2124 frames
23.3 * 90 seconds == 2097 frames

2124 - 2097 == 27, or just over 1 second. Unless you're filming a really dialogue-heavy movie, this just won't be too much of a problem. While you're editing you frequently have to spend hours tweaking the audio anyway; this will just be one more step.

Edit: you can always keep a stopwatch going on set, and when the take is over or gets close to 90 seconds, you can run in and do a tail slate. With a head and tail slate, you can just stretch the audio to fit.

- ben

[Juan M. M. Fiebelkorn]

Ben,
Not to offend you but you are not getting it with the color correction (CC).Your images are too green/yellow.
I have a couple of images corrected that I can send you.
Can I?
I've also corrected Obin's image posted by Jason and it looks real nice!!!

[Les Dit]

Ben, The frame rate is probably a software problem. Instead of taking a frame after every 'N milliseconds' they should be tying it to an absolute timer, taking a frame on every tick of that timer. That way the most problem they would run into would be jitter in the individual frame times, but no long term time error.

It they can pack the 10 bits into the USB with the full frame rate, that would really be great news. Maybe they will be doing some mild compression?

-Les

[Matthew Miller]

<<<-- Originally posted by Juan M. M. Fiebelkorn : My bussiness partner has around 25 years of experience as DP.He has made more than 5,000 commercials from here and the rest of the world, and agrees with me in what I'm saying. -->>>

Since we all seem to be in the spirit of touting our expertise and bickering over mathematical equations, I figured Juan might have missed a chance to truly express the magnitude of accomplishment that his DP business partner has enjoyed in his many years of service to the film industry.

5,000 commercials divided by 25 years = 200 commercials per year.
5000/25=200 right?

And there are only 365 days in a year, though 366 on leap years, which occur every four years.
25 years divided by every 4th year being a leap year = 6.25 leap years, or 6.25 additional days.
So... 365 days in a year multiplied by 25 years plus 6.25 days for leap years.... equals....
(365x25)+6.25= 9131.25 days

So less than 10,000 days have gone by in 25 years, yet the DP Business Partner of Juan M. M. Fiebelkorn has managed to make 5,000 commercials.

9131.25 / 5000 = 1.82625

and we all agree that 1.82625 is less than 2... right?

So DP of Fiebelkorn managed to CONSISTENTLY make a commercial every two days for 25 years.....
......
.....
....

And it took God six days to make the earth.

[Ben Syverson]

Juan, I am "getting it." Color correction is a very personal thing -- you may want the outside shot I just posted to be less green, but I like it with the balance it has. In fact, I like it a little bit bluer too.

The indoor shots I posted before were taken without a #80 blue filter. Without higher than 8 bit images, you cannot white-balance in post. You can adjust the gamma a little bit and you can push the saturation a little bit, but white balance is the equivalent to adjusting the input whites in Levels. That will give you banding all over the place as soon as you try to push the saturation.

Wait for the 10/12 bit software from Sumix. Then I'll be able to push these images to where you think they "should" be. Until then, they aren't going to have perfect white balance and color correction like some video shot by the government.

@Matthew: "So DP of Fiebelkorn managed to CONSISTENTLY make a commercial every two days for 25 years....."

<chuckles> I'm sure you're DP is great, Juan. You just have a way of telling other people they're doing things wrong. "Don't color correct that way." "Don't shoot with that lens." "You have to do it this way."

I'm just saying... if you want to show us how to really do it, order a camera and start posting images.

-ben

[Juan M. M. Fiebelkorn]

ok, I just made a mistake and put a zero while being in a hurry,
SORRY.
About white balance I have the imges and I want to send them to you Ben.
The easy way I use is to measure RGB values for a white zone, and balance according to that.
I don't really understand why I do annoy you so much.
mail me fiebelk at hotmail dot com


So tell me.
Must I stop making posts?
Do I have to go somewhere else cause I'm really stupid?
Everything I say is really a nonsense?

Cause I'm starting to think that.

[Matthew Miller]

Juan,
I think we just have too many people here who like to talk about how smart they are and wave around a list of what they have done in their life as some kind of certification of their right to be on this thread.
It is as if there is an unspoken contest to see who can contribute the most posts to the Alternate Imaging Methods. The winner gets recognized by the other members of this message board as "Grand DV Master" and all must bow to him.

Ben said it right. Anyone who wants to show us all how to do it right is gonna have to buy a camera and start posting some results. Otherwise negotiate with Obin or Ben how much you'll have to pay one of them to do exactly what you want them to with their own equipment.

Ben never said anybody was annoying him. Let's try to keep it that way so he keeps sharing his results.

[Rai Orz]

<<<-- Originally posted by Wayne Morellini: I only suggested the version without the GG to increase the image brightness in low light situations, using a seperate 35mm lense. As far as I can figure out the aperature ratio tells you how well lit every piont is over a set area. 35mm has, 4 times more area than 2/3"???, so concentrating a good F1.0 slr lense onto 2/3' chip (though yours is 1/2") would give two stops more light than a F1 2/3" lense. Is that right?
-->>>

It sound you will have more than 100% light. But this is definitely not possible. Every lense system loses light. But the point is the front side diameter (in relation to the projection diameter) of the lenses. A lense-system with 2% light loss but with 70mm diameter catches mutch more light as a 20mm lense with the same 2% loss. Just a look to the size of a lense-system can bring a idea how bright the image will be.

Wayne, can you contact me via email (reg. cases and others) ?

[Wayne Morellini]

This is the common missunderstanding people has about what I'm saying, what I am saying is the light from a large film area is condensed down to a smaller film area=brighter image, no light production. In the other thread I had, people kept on stupidly arguing about interpreting words to mean producing light, rather than gathering more light to a smaller area, they got caught up on words (crossing the I's and dotting the T's ;) rather then what was meant, (like English majors trying to argue history from the gramma they spoke, instead of what was meant). Which I know your not doing.

[Obin Olson]

Jason, I would say that gain makes things worse but is not the main issue..it's just when you shoot a blownout light it streaks! the less gain the less streak but it's still in the darks...rolling shutter is fine at 40mhz and above IMHO...it's the streaks that worry me..I would not buy a camera with usb or firewire..not enough datarate when you compare with CameraLink

[Wayne Morellini]

About Summix, "Steve I" and Summix actually started this whole custom HD camera thing here, Obin decided to go with SI instead. They are very interestred in doing stuff with us and we have old announcements they have made on the other thread about everything discussed here (apart from the new software 12-10-12 improvements and firewire). I wish everybody would read the threads rather than rehashing information and questions. (Sorry, I hate to see unanswered questions to new people (and even suggestions) because people consider them old news, and the rightful recognition drifting away).

Re-edit: I forgot to mention the use of Firewire B is dissapionting, unless it can gaurantee 100MB/s (after overheads) like we can get out of Gigabit Ethernet. Can you find out if itcan Ben, thanks?

Competitions, I got nothing to compete about in production, or camera technology, this is the real training ground, I read and learn, and rarely get involved in production/post production issues. I contribute to the technical side because I am aware of many of the issues (from experience and learning) and I want to see the best solutions (the most appropriate low cost quality solutions). Not really a competition, but sometimes some people want to fling it like they know it, and have to be enlightened.

Now Jaun, don't get offened by people, you just come accross differently from what you intended in English. When you read it sometimes you missunderstand it differently than is intended. It is just the way you speak is differrent from normal english and the gramma is different. I also have a bit of this problem (my sentences back to front and complex), this has made it diificult for you and Obin to read my stuff aswell.


For everybody:

Now the other problem is competitions, and being unquestionable etc, rather than discussing and working to an understanding.

Don't get offended by other's niave questions, probing, and suggestions, it is just part of communicating and coming to an understanding of what is happening, and sometimes the people recieving them are actually mistaken, despite what they think, and I am allready aware of my own weaknesses.

So peace to everyone, lets discuss and continue on.


Frank Roberts, made a good post. The systems he sugested are interesting, and the machine vision section of Edmund optics is interesting, with a cheap 2/3inch HD camera (hadn't researched that properlly yet).

Somebody with more practical knowledge please help and encourage him.

Thanks

Wayne.

[Wayne Morellini]

Well I know some people ignore my posts, that is why they keep asking questions allready answered. But I posted them because they are very benefical to the propper developement of the project, and they have taken a lot of research hours, so take advantage of them.

Now I haven't contacted Apple, VIA or X number of other companies. A major multimedia event software firm, Cinelerra, case manufactures, adpator manufacturers, or Sumix (as was suggested), because there is too much to do to get the project from adhock (red herring) to mature pro project.

Now the volunteer resreachers post I did yesterday is needed. Sure everything is adhock at the moment, that is normal, but it can't stay that way, otherwise everything people have criticised over at cinematography forums (and even that camcorderinfo forum) about this project will be true. Obin, I think you know what I mean from recent events, you want good software and cases now, and good workflow, I agree (though it will still take some time). I have been talking to the Rob's and Steve I about big corporate sponsorship, funding and development, and even past that.

Now, I wanted to do a few political, advocacy and fundraising things (aswell as some hardware development for cases etc) for the project (I suppose it is because I am feeling good at the moment) but I need help, it is too much (just keeping up with the threads).

We can rock everybodies world, but past the experimental cameras of Obin and Ben, we have to get seriouse, and structured, if we want an effortless camera system this year. I don't volunteer for anything (for health and commitment reasons and because I will lend up doing too much without volunteering anyway). But if you need a project manager, or representative, I can volunteer some time (with some assistance).

I have a good grasp of techncial and business issues. I used to run my own technical advocacy website, and am planning a social activism site (to use the camera for). Have been around people developing expensive hardware projects (processors, and computer devices). I also have researched into new designs and concepts for the following devices: PDA's, computers, laptops, case formfactors, handheld/watch/projected game systems, 3D/volumetric game portable game systems, and projected displays, bit map to serial augmented display, text entry systems, keyboard systems, game controllers, bus networks (abstract), Virtual OS system (with a improved versions of nearly every system in the OS and computer, including processer and 3D pixel system), low light loss SLR adaptor (intial research on hold for this project), camera case systems, hybrid HD acquistion etc. Generally every aspect of anything I'm researching, with new origional concepts and practical solutions for most things (a number replicated independently in the industry, i.e the simple Palm Pilot formfactor). Unfortunately hardware ussually requires much too much money.

This project has potential to be sold as a good marketing and business opportunity.

I also have to start work on my OS again sometime (that was to fund many of the hardware projects) but this camera is the sort of social activism project I like.

So you can appreciate why I don't like bothering with competitions, and concentrate on practical "long term" marketable solutions.

(About the competitions, if your good, you got my appreciation right away, unless you need to convince me (prove it by discussion or action), otherwise you can still get my respect. It's a good policy to adopt, rather than getting offended and being unquestionable. Understand what each other is really saying (their meaning, and piont of veiw or opinion), don't expect too much, but appreciate that you may not be able to agree, or have two different opinions or pionts of veiw (agree to disagree).


Thanks

Wayne.