Home Made HD Cinema Cameras - Technical Discussion

[Régine Weinberg]

this card I did mention in the old thread Olbin deja...2005 0r 2004
it is SCSI and the drives need a lot of power hm,
but the express cards works, there was one at the time
with dual cameralink and scsi bypassing
the PC literally only writing data to disk

[Wayne Morellini]

Originally Posted by Ronald Biese

this card I did mention in the old thread Olbin deja...2005 0r 2004
it is SCSI and the drives need a lot of power hm,
but the express cards works, there was one at the time
with dual cameralink and scsi bypassing
the PC literally only writing data to disk

Yes I forgot to mention that is where I found it, and probably why you have been so peeved lately because people are not listening, otherwise they would have seen it. You post so good links, but only a few of us takes the time to looks through links. I did not see it because I previously stopped going to that group.

If it was only a realistic price, like $500 with ATA drive support (that will equal at least 100MB/s sustained, and shortly probably 140-150MB/s, 300MB/s sustained for three drives. And such a device could be put into a Gige box easily, because only a back display channel is used through PCI bus (the camera data goes directly to drive without goign through computer) or even have display directly from the box. Eventually a device like that could fit into laptop PCIE card fro 250Mb/s display channel, but that will take a while and probably custom ASCI to get it to fit (but the Ambarella could also do it and compress the display channel through a ordinary laptop PCCARD bus). But fro now this is what we are stuck with.

[Steve Nordhauser]

Wayne,
The last time I checked, there was only a SCSI version, the PCB (single piece price) was $3500 and the software (required) was $2500. IO Industries is a very good company but they have a military/industrial leaning that is not financially in line with Indie needs. That said, I am working with one creative person who is using our older 3170 on an IO and just received the SI-1920HD. He may be making systems available for people. It might be a good answer for high frame rate recording.
Regards,
Steve

[Wayne Morellini]

Sorry to hear that about the software. I had done enquiries and even found that to record to disk seamlessly for many hours you need an expensive version of xcap for Epix cards. Their current packs of camera/card prices are around $1K now.

It is a shame that nobody does a pro quality video recording software solution based on the machine vision software (but Lite version, stripped of all extra irrelevant machine vision features) that can be optional ship instead with frame grabber/cameras. Imagine how many extra sales they could have got as an $1K+PC alternative to video people. Telling people they can only record to main memory at system prices similar to HDV is not appealing enough.

Good to see the new camera, and wish you luck with it. Any plans for Box cameras based on the new Foveon, or new 720p Altasens?


Thanks

Wayne.

[Steve Nordhauser]

Wayne,
Actually, it was this discussion that caused us to write our own software package. The machine vision stuff is the wrong starting point. 3x3 deBayer, no log conversion, all the tools are wrong. We did the uncompressed recording in a cinema-based format and discovered that unless you have huge resources, you can't edit uncompressed. That is how we ended up with the Cineform workflow.

We will be offering the camera head & software solution for $12Kish for people who want to roll their own. 720p is possible for a next product but it seems everyone is pushing higher, not lower in resolution. I'm sure we would never reach the price you mentioned.
Regards,
Steve

[Wayne Morellini]

It is all relevant, the resources are here shortly. Memory, Disk space (750GB pr disk) and processing power (CELL, AMD/Clearspeed, DX10/11, or future dual or quad with low powered chips) is within the realm of higher then 1080p. When we started, a couple of years ago, the sort of PC that we needed to do this workflow would be the most cutting edge (realising that Cinelerra was being used for uncompressed HDSDI 1080p 4:4:4 movies at the time). I am aiming to contact David about something that could feasibly allow cineform for much else then a thousand dollars). But for the low end 720p lossless Bayer is enough and doable with the latest computers.

So given a debayered processed file I think Cinelerra can edit it at 720. It doesn't matter, the whole DIY camera stuff is in it's twilight for the moment. Though I still hope for something less expensive alternative to HDV that the most new, or low end, videographer can run around with.

Thanks

Wayne.

[Régine Weinberg]

We all do know the P III will be out November.
What vavour of Linux Sony wiil put on the disk,
who knows? If Cinarela will compile who knows ?
I will talk to Dynebolic, Jaromil to have it run on The PIII
as open Mosix will build an cluster quite easy and
without to be a network guru. To have 2 or three Cell
in a cluster config will make Cinarela fly.
Could be a Dream Machine

[Steve Nordhauser]

Wayne, don't think of it as twilight for DIY, think of it as evolution. First, I think DIY was a catalyst for the new low cost HD and camcorder releases, certainly for ours. Also, I'm old enough to have been a subscriber to Byte magazine from issue #1 in the 70's. I built a number of computers from scratch including one with paper tape and cassette storage and another with 8" SSSD drives (90K on each floppy!). I thought the cheap z80 CP/M machines, followed by PCs would kill the electronics hobbyist.

It didn't kill the hobby but it changed it. First there was a flock of home programmers - you didn't need a masters in EE or CS to program them. Then came the microcontrollers. Any clever person could breadboard a software driven motor speed controller. Not many people go to Radio Shack anymore for parts for serious projects but Digikey and Jameco still sell a lot to hobbyists.

So, maybe DIY will transform - inexpensive time lapse, stereo, lens adapting, etreme low cost, something. In the end, everyone has benefited from the DIY in HD.
Steve

Originally Posted by Wayne Morellini

So given a debayered processed file I think Cinelerra can edit it at 720. It doesn't matter, the whole DIY camera stuff is in it's twilight for the moment. Though I still hope for something less expensive alternative to HDV that the most new, or low end, videographer can run around with.

Thanks

Wayne.

[Wayne Morellini]

The Past:

Yes, "Byte" and "Creative Computing" those two stick out, I know there were a couple of other really good ones, but the PC industry killed them all, except for a few of the best UK ones, and the Australian Personal Computer (Aus version of Personal Computer World) (I don't know about Byte, haven't seen one for many years here).

Those CP/M etc machines did not kill the Electronics Hobbyist, but they and the ones that followed definitely did kill the computer kit industry (some of those things had such an array of components on their boards, they looked like a work or art). The only non major brand name I am still aware around of, is Super Micro, or something like that (Acorn computer might still be around). In the end it was about owning a computer rather then spending long long hours putting them together.


The Present DIY:

I am not sad at the demise of DIY for it's own sake, it was always about cheap high quality cameras, that still has not happened. I recently located a HD webcam (about $146) that claimed 24fps at 1.3mp resolution. I have been communicating with them for a development kit, and it turns out that the 24fps was a misprint (it was on all the pages, so I though it was genuine) it should be 15fps :(. Major disappointment, it might have been little better than a consumer HDV cam in terms of noise and latitude, but the possibilities for higher quality codec recording was definitely there (I did not get around to verifying if it had RAW). I have located a number of webcam related DIY webcam astro pages, and RAW and more bits per pixel should definitely be possible.


Some fruit:

But, the DIY projects have born some fruit, I feel. Since the projects were announced the market has taken unprecedented changes, that have stolen away our members as well, I feel. We got cmos HDV HC1, at such quality it has been alleged as stealing FX1 sales (a friend of mine was even looking at buying it instead of a FX1 because of size, but I was advising because of latitude, before the HC1 got canned). Although I don't think that XDCAM HD was because of us, at least it is not a 25mb/s pro version of the consumer HDV, as Sony did with the DVCAM version of the Mini-DV. I think the 18mb/s AVCHD Mb/s inter H264 may well be a reaction. Quiet frankly, under 9mb/s is what you would have expected for this (from past trends) I think the problems with motion/complex scenes, low light noise affects of compression, blu-ray and HD-DVD also has been a large part). I think that the 50mb/s hi-profile 10 bit 4:2:2 h264 version from Panasonic being rumoured to go into the next HVX200 camera is also a reaction (high profile h264 goes into many hundreds of megabits, 12bit 4:4:4 and lossless, for intent of professional work and editing). This does not even get into your and Drake's cameras. There are probably others that I am forgetting, where the industry might be hedging it's bets, and heading DIY off at the past.

The truth of the matter, is that if something with encoding quality/bitrate as good as XDCAM HD (or AVCHD) had been available instead of the HC1/JVC HD1, I would probably have gone with it instead of DIY. If it had been as good as the 10 bit 50mb/s Intra Frame h264 HD, I definitely would have. That would have been nearly ideal (I am not saying they should have introduced a h264 codec, but something with similar quality where ever upto 200Mb/s DV like codec, or 50Mbs+ Mpeg2 inter). I definitely think that the use and reliance on Mpeg2 for HDV (which was around 10 years old at that stage, where as Mpeg4, introduced after Mpeg3 was rolled, was getting old) was a mistake, as much as restricting the resolution to interlaced 1440 lines, 4:2:0 codec and data rate to 25mb/s. They should have had Mpeg4 codec, or a quick migration path to one, at those tape speeds, as Mpeg4 technology in consumer, and then in cinema camera, had been available for some time.


The Future: Cineform, H264 and Ambarella.

I see these new H264 formats as at least one step down from your camera, with lossless being the best. What I want to see now, is something of the quality of cineform bayer to replace the h264 codec. People might feel this is threatening, but what people don't realise, is that the quality of encoding is not the real underlying discriminator for the future, but the resolution you work at. It costs money to push the real physical limits to get a sensor to perform well at high resolution etc. So the market can now be stratified as: 720p to 1080p for prosumer, 8Mp (yields a better 108Op image) 16mp and 32Mp for cinema and future TV.

I have contacted David about the h264 Ambarella camera control/codec chip, as it is a large array of Sparc RISC microprocessors, opening it up to the possibility (though I am yet to obtain information to verify it, but I think it would be likely) of being programmed with any codec, like the cineform bayer codec, with the extra processing power taking up the slack of lack of cineform specific features. The chip works at over a hundred Mw for the low cost (from $25) consumer versions, and under 1W pro version. As you can see, if such a thing could work, it would mean that all pocket cameras, all camcorder, all prosumer cameras, and all industrial box cameras could use something like this, as well as your present Altasens camera. Yes, I imagine it might not be as simple as that, but I am yet to obtain a development kit in order to research it (I am looking at it for the development of other products). I don't think David was much interested though. But for your purposes in other fields even h264 is interesting.

I would be interested in your observations on these comments Steve.


Thanks

Wayne.

[Wayne Morellini]

For all that have been here, you will be familiar with me being interested in the general video processing potential of future GPU structures. Here is the latest on the next generation of ATI GPU processing unit:
http://www.theinquirer.net/?article=31862

It is following a year after the one that shipped in the Xbox 360, similar to what happened between the original XBOX/s GPU and a PC equivalent.

[Steve Nordhauser]

Wayne, there are many discriminators for the 'best' camera. Certainly resolution but also dynamic range, sensitivity, bandwidth, ease of editing, weight, cost, lens compatibility, ergonomics, upgradability, and many others. The importance of each depends on the user and the application. If you are backpacking your equipment for a documentary, weight is more important. If you are making a movie with your allowance money, cost is foremost.

On the processing chips, volume is usually a barrier. I have tried to incorporate DSC chips from a number of companies into low cost video cameras. 10K units/year is the minimum to get them to open the door and talk, 100K is what it takes to get to a deal. The issue is that they can only properly support a certain number of design-ins and a single sale takes the same support (almost) as a 100K sale. That is why the development kits are usually $20K+. Not to make money but as a barrier to get out the riff-raff.

This is why we went with more conventional PC technology. The CPU speeds are now fast enough for compression and GUI. Lower volume sales are acceptable to the vendors. Development is also easier.

Steve

Originally Posted by Wayne Morellini

I see these new H264 formats as at least one step down from your camera, with lossless being the best. What I want to see now, is something of the quality of cineform bayer to replace the h264 codec. People might feel this is threatening, but what people don't realise, is that the quality of encoding is not the real underlying discriminator for the future, but the resolution you work at. It costs money to push the real physical limits to get a sensor to perform well at high resolution etc. So the market can now be stratified as: 720p to 1080p for prosumer, 8Mp (yields a better 108Op image) 16mp and 32Mp for cinema and future TV.

I have contacted David about the h264 Ambarella camera control/codec chip, as it is a large array of Sparc RISC microprocessors, opening it up to the possibility (though I am yet to obtain information to verify it, but I think it would be likely) of being programmed with any codec, like the cineform bayer codec, with the extra processing power taking up the slack of lack of cineform specific features. The chip works at over a hundred Mw for the low cost (from $25) consumer versions, and under 1W pro version. As you can see, if such a thing could work, it would mean that all pocket cameras, all camcorder, all prosumer cameras, and all industrial box cameras could use something like this, as well as your present Altasens camera. Yes, I imagine it might not be as simple as that, but I am yet to obtain a development kit in order to research it (I am looking at it for the development of other products). I don't think David was much interested though. But for your purposes in other fields even h264 is interesting.

I would be interested in your observations on these comments Steve.


Thanks

Wayne.

[Wayne Morellini]

Originally Posted by Steve Nordhauser

Wayne, there are many discriminators for the 'best' camera. Certainly resolution but also dynamic range, sensitivity, bandwidth,

Of course, but that is what I meant by getting the higher resolution pixel to perform well, and basically also four times the pixels adds four times the bandwidth and storage to the workflow. With the same lens, encoder and equipment, with an optimal performing pixel (harder at higher resolutions) I think it is the resolution that is left as the difference, and the the resolution of the delivery format you are working it for.

On the processing chips, volume is usually a barrier. I have tried to incorporate DSC chips from a number of companies into low cost video cameras. 10K units/year is the minimum to get them to open the door and talk, 100K is what it takes to get to a deal. The issue is that they can only properly support a certain number of design-ins and a single sale takes the same support (almost) as a 100K sale. That is why the development kits are usually $20K+. Not to make money but as a barrier to get out the riff-raff.

;) (Some companies are better than others).

I think the Ambarella chip may be well thought out and flexible enough to make integration into various designs a minimum of fuss, but something still done by SI rather than them. Even if the chip was ten times it's price, say $250, it is still a lot less then a PC, or designing a FPGA system. This is probably the best re-programmable processing deal that I know of on the market presently. If it holds to be a mainly re-programmable software system (without hardwired processing stages) the programming development costs would also be similar to a PC. There is also one other market possibility, if they don't support Machine Vision directly, there is a potential market with other machine vision suppliers, for applications and drivers that do.

Still, you see that eventually, a cheaper camera is possible in future.

Have a good day Steve.

Wayne.

[Wayne Morellini]

For the benefit of the members so they may look up different sensor information:

I have found a digital camera IC site, with a master list of companies for controllers, processor and sensors:
http://www.perfectdisplay.com/id45.html
http://www.perfectdisplay.com/zic_im...ng_999999.html

This is a master list of IC companies (search for "sensors", and you should see "imaging", or "image", besides it:
http://www.eworld.ru/ewg/icweb.htm

A list of sensors, would include all non image sensors, so some aren't of use.
http://biz.cnnet.com/archives/company/523/Sensors.html


I include a list of the companies and links on the sensor list, as the site does not look current and probably will be removed:

Agilent Technologies, ICD
Altasens Inc.
Advanced Analogic Technologies
Arima Computer Corporation
Aselsan, Inc.
Atmel
Axis Communications
Canesta
Core Logic Inc.
Dalsa
Dialog Semiconductor
Displaytech
Elecvision
Endpoints Technology Corp.
ESS Technology
Foveon
Freescale Semiconductor
FujiFilm Microdevices Ltd.
IC Media Corp.
iEagle Microsystems Limited
IBM Microelectronics
Irvine Sensors
Kodak Image Sensor Solutions
Lite-On Semiconductor Corp.
Magnachip Semiconductor Ltd.
Megachips LSI Solutions
Micron Technology, Inc.
MobilEye
Mosart Semiconductor
MtekVision
National Semiconductor
OmniVision Technologies, Inc.
Panasonic
Photonfocus AG
Pixart Imaging
PIXIM Inc.
Seiko Epson Corp.
Sharp Microelectronics
Sony
Soundpix
STMicroelectronics
Sunplus Technologies
Symagery Microsystems
Texas Instruments
Toshiba
Tower Semiconductor
TransChip Inc.
Winbond Electronics Corp.
Wolfson Microelectronics
ZMD America
Zoran


http://www.agilent.com/
http://www.altasens.com/
http://www.perfectdisplay.com/id359.html
http://www.arima.com/
http://www.aselsan.com/
http://www.atmel.com/
http://www.axis.com/
http://www.canesta.com/
http://www.corelogic.co.kr/
http://www.dalsasemi.com/
http://www.diasemi.com/
http://www.displaytech.com/
http://www.elecvision.com/
http://www.endpoints.com/
http://www.esstech.com/
http://www.foveon.com/
http://www.freescale.com/
http://www.fujifilm.co.jp/
http://www.perfectdisplay.com/id420.html
http://www.ieagle.com.hk/
http://www.chips.ibm.com/
http://www.irvine-sensors.com/
http://www.kodak.com/
http://www.liteon-semi.com/
http://www.magnachip.com/
http://www.megachips.co.jp/
http://www.micron.com/
http://www.mobileye.com/
http://www.mosart.com.tw/
http://www.mtekvision.com/
http://www.nsc.com/
http://www.ovt.com/
http://www.panasonic.co.jp/
http://www.photonfocus.com/
http://www.pixart.com.tw/
http://www.pixim.com/
http://www.epson.co.jp/
http://www.sharpmeg.com/
http://www.sony.com/businesssolutions
http://www.soundpix.com/
http://www.st.com/
http://www.sunplus.com.tw/
http://www.symagery.com/
http://www.ti.com/
http://www.toshiba.com/
http://www.towersemi.com/
http://www.transchip.com/
http://www.winbond.com.tw/
http://www.wolfsonmicro.com/
http://www.zmda.com/
http://www.zoran.com/

[Wayne Morellini]

New Sensors:

This is the Pixim company, I have been supposed to research this stuff, but have not been able to do it. I include this is a hope that somebody else might have the time to look this up and research it for us. The Sensor is supposed to have extended slope ability (for extremely high latitude) and sounds like those smalsensor stuff. Cheap, and a list of camera manufacturers, lens manufacturers and other items included:

Sensors:
http://www.pixim.com/html/prod_matrix.htm
http://www.pixim.com/html/prod_chipset.htm
http://www.pixim.com/html/prod_chipset_orca.htm
http://www.pixim.com/html/prod_chipset_dyna.htm

Cameras:
http://www.pixim.com/html/prod_cameras.htm

Camera Development:
http://www.pixim.com/html/prod_chipset_cdk.htm


Lens:
http://www.pixim.com/html/prod_camera_des_res_len.htm

Camera design resources;
http://www.pixim.com/html/prod_camera_des_res.htm

Industry partners:
http://www.pixim.com/html/prod_industry_partners.htm

--
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/

[Wayne Morellini]

This is a closing summary of the DIY projects, and the quality cheap alternatives I have found to do them. I have spent much time researching things in the background, and probably have hundreds of links still that have not been researched, into interesting things, parts and possibilities, if anybody is interested? It is a bit too much for me, and I need to get on and do other things. The new, interesting, commercial alternative cameras are listed at the end.
----------------------------------

I have been doing some research into the different ways cameras could be done cheaply, and I have come up with some alternatives:


Still, Mobile, Flash Camcorders, webcams
I have come across information that makes me believe that some of those Digital still cameras, mock solid state camcorders, and phones, can do upto 12 bit raw bayer. Sure you will get the quality of something like a consumer/prosumer HDV camcorder, but uncompressed, and for many people that is all they need. At the moment HD from webcams is limited by frame rate, but inevitably webcams will get there with still cameras, phones and solid state camcorders.

Work undone: research on all cameras, sensors, and codec/control chips.


Webcam Modifications and custom Linux drivers already on the road
What will help is that there have been a history of writing custom drivers for webcams for Linux, and the Astronomy community has been modifying webcams for Astronomy work, including RAW uncompressed (I forget but I imagine higher then 8 bits) changing hardware, something dealing with sensor masks, and changing sensors for better ones ;).

This thread has links to the astro webcam modification community:
http://www.dvinfo.net/conf/showthread.php?t=63837


Security Camera
Can be relatively cheap with better quality sensors and compressed data. Elphel have a model people are testing, but the Elphel 423 will have better support for the Ogg Theodore video codec, so I expect might be a descent camera, not cineform or RAW uncompressed, but cheap too. There are other security cameras.

Work undone: research on all cameras, sensors, and codec/control chips.



cheap HD pixel shift
There are PC mainboards with multiple SD camera inputs, you configure a 3 chip camera with pixel shift, you get HD (ask Andromeda how it works).

Work to be done: Still working on it.


radical alternative MiniDV to HD
Looks like a washout. I did work out a possible cheap system to get HD from a MiniDV camera, but the DV codec does not record enough data, and the pixels are too imprecise, so not even DVCPRO50 will do. Priority method. Maybe it can be done one day.


Recording better quality from HDV etc HD cameras
You can record from component, I did find a card with 5:1 compression around DVCPROHD resolution around $300, mentioned earlier. Uncompressed cards I think now start around $1K, and the $300 card I mentioned, I imagine it might be able to be reprogrammed to do uncompressed. Their is an alternative, there are PC mainboards with multiple camera inputs (mentioned with the pixel shift above) theoretically they should be able to record HD component (I have confirmation from one chipset manufacturer that this was an aim for their chipset).

http://www.dvinfo.net/conf/showthread.php?t=61343


As a side issue, I am trying to find out what modes and video formats of transmission come over firewire or USB2.

http://www.dvinfo.net/conf/showthread.php?t=68054


Now, there was a thread somewhere, that tested the video output (or did that include firewire too) of a camera when it was in memory mode (for still shooting). Suddenly the image was of better quality, but no conclusive testing was done to see if it was uncompressed from memory. If you search under my name for threads with "memory mode" and "uncompressed", you should find it easy to find. The issue is, what happens to component/firewire/usb output on HD cameras in memory mode.

http://www.dvinfo.net/conf/showthread.php?t=51632
http://www.dvinfo.net/conf/showthread.php?t=51795


Undone work: as above, testing etc, finding cheaper uncompressed component cards etc.


Cheap machine vision frame grabber and camera combination
Quality Machine vision setups are cheaper than broadcast setups, that is why they are still important. I noticed that Epix has frame grabber and camera packages starting from less than $1K, but these have limited software. I have communicated with Epix, and if you pay for the more expensive software you apparently will be able to seamlessly record hours of uncompressed Raw footage reliably (advice talk to them and confirm it will do what you want, I think test can be arranged, but I forget if that was them). This was the Xcap software people complained about, makes you wonder why we bothered. The new PCI express frame grabbers are apparently the ones that can do it, pixel packing and buffering is used (along with simple recording software). Direct to disk recording software has been described to me, probably by another company, that the "tricks" to writing successful direct to disk recording is, simple, actually I have some of the advice in an email somewhere, or was that by phone). I wish I was upto scratch with realtime embedded PC programming I would know what he meant ;).

Yes, Gige/USB2 cameras (though USB has it's own problems with wasting away processing power, but I wonder what happens with dual core systems nowadays). Sumix is fruitless, don't even reply to email anymore. I pity them, if they ever turn up and find everybody has gone because those people thought it was dead, and virtually nobody is left to buy their marvel. Some cameras out there have FPGA and embedded processors, enough to set up record to USB/Gige caddy with some simple compression like that which I was planing? Some frame grabbers have FPGA too.

Work Undone: Looking through all industrial Camera Link/USB/GigE/frame grabbers/capture software to find the cheapest quality cameras/frame grabbers/software and system combinations (links to lists in previous posts). I would like to see Foveon in some cheaper cameras.


Existing used free software for Hollywood
Another, often neglected avenue, is the open Sourced Cinelerra, which has been used in Hollywood, and to capture motion pictures for years. It is a wonder that nobody has talked with that community about machine vision camera support. I did email them (probably twice) but received no reply, so I usually say, stuff them, to people like that, I am not trying to help people to get ignored, where ever rude idiots or not, so that was the end of that. The problem is that I think it is limited to HDSDI/component cameras, meaning high cost solutions anyway.

http://www.heroinewarrior.com/cinelerra.php3
http://sourceforge.net/projects/heroines/
http://www.heroinewarrior.com/cinelerra/cinelerra.html


Local HDSDI RECORDER
Of course, my current favorite hero, is Keith Wakeham (well where did you go Rai) who is bravely doing his own FPGA HDSDI to disk recorder unit. At the moment it is HDSDI but it could go to component (say for current, and future, HD camcorders with component) but component is a hairy (fuzzy) beast that limits it getting done, but bayer machine vision is possible. The price that Keith hopes for would make it an attractive alternative, but the price of HDSDI cameras are not. So, for the moment, guys with JVC HD250 and Canon HDXL should be kicking with joy.

http://www.engr.mun.ca/~wakeham/index.html


Makes me wonder why nobody gets a camera control chip and makes a direct to disk from USB/GIGE from that. Ambarella is a descent scheme, but they only seem to be interested in real big volume, and not talking with less, as with most.