stuckat1

On 4/14/2015 at 6:15 PM, Julian said:

Or something like this

Olympus VF-4 external viewfinder. 2.36 million dots, tiltable and very compact. costs only like €200. Would be cool to see a small/cheap optional evf like that which works over hdmi. Shame it's impossible to power it over hdmi.

This is a really nice idea.

I've done some research on the feasibility.  Unfortunately, the view finder is actually an EPSON LCD with 1024 x 768 in resolution 4:3.  (1024*768*3 = 2.36m, 3 is for R+G+B) using a LVDS interface.  The camera outputs HDMI or composite.  Because the HDMI is FHD (1920x1080) 16:9, one will need a LVDS to HDMI driver WITH A scaler to downscale.  Unfortunately, while there are plenty of drivers, most do not downscale.  For those that upscale/downscale none that I could find would actually adjust for different aspect ratios.  You could theoretically replace the Epson LCD inside the viewfinder with a FHD one, but all those are a little larger but only do black and white video.  Because of the larger LCD size, the optics might need to be upgraded or readjusted.

The other option is to use composite video to either drive the viewfinder or a small external LCD monitor, say 4" or smaller.  The problem with this is that composite can only generate up to 480i.  This is probably not good to judge exposure or more importantly critical focus.

Another option is to use the same 5" screen the  BM Video Assist uses or a screen from a high resolution cell phone. Those screens use DSI protocol over LVDS.  Right now, there are no boards that do HDMI to DSI for sale.  There a two boards but they are not for sale ... yet.

One can only keep dreaming.

On 2/14/2016 at 3:54 PM, blafarm said:

Not sure you are speaking from experience.  However, as the owner of 8-Android devices (combination of phones and tablets) and 11-OTG cables from various manufacturers, I can tell you that not every USB device, works with every OTG cable, on every Android platform.  And in fact, I currently can't get any OTG cable to work on my Galaxy Note Pro 12.2 (which is a USB 3-based device).

It's actually quite vexing, and it becomes even more complicated when the power consumption requirements of the attached USB device are factored into the equation (which would certainly be the case with this scenario).  And this performance inconsistency is not necessarily resolved through the use of OTG cables that offer a separate Micro USB power leads.  

Unfortunately, Android was never developed with any emphasis on the ease of connecting external USB-based devices -- and years later, it is still an afterthought.  For this reason, Windows-based tablets will likely have a leg-up in this realm.

Here are two one, but there are others.  You just have to look for hybrid devices that function as both tablets and external screens ...

Wacom Cintiq Android Tablet

Acer Aspire Android Desktop

But the bigger fundamental issue, which you already mentioned, is that tablet screens were never manufactured to display the frame rates that emanate from our cameras.  They were designed as a closed ecosystem -- optimized for one function (and that function is clearly not video assist).  Whether you use a capture device -- or jack-into a tablet that has an HDMI input, frame rates like 23.976, 29.97, 59.94 will either cause unacceptable motion artifacts -- or it won't be displayed at all.

 

 

Great comments!!!

I do think all keyboards and mice using standard HID configurations that don't draw too much juice should work with all Android tablets (that have intact USB stack implementation) with USB host support and with a OTG cable.  I imagine its completely possible that devices get shipped with little or no USB profiles other than mass storage.  This is probably most likely for cheaper devices, under $100.

Wacom and Acer Aspire devices are very interesting.  Never seen them before.

The Acer Aspire is so physically large and inexpensive that I think they just literally built an internal video switcher inside so the video isn't being processed by the OS.  I watched a promotional Acer Aspire video that shows that the OS screen just vanishes when a HDMI cable is plugged in.  A switcher probably only costs $5 to manufacture but you cannot record video from the source.

Wacom is an interesting device but it does cost $800+.  Wacom obviously designed special hardware to display the video feed inside its apps.  This could definitely record video but the problem is that Wacom probably has no interest in making those APIs public.  This means no third party support. Also this device is 13.3"!!!

Also, the Wacom is designed to communicate with a PC using DCC, so it probably supports one or two display modes/resolutions.  The tablet and PC negotiates bidirectionally how to display the best resolution/frequency.   When you plug a monitor to a camera, the camera doesn't negotiate. It forces a resolution and refresh so if your HDMI recorder device can't support it, your S.O.L.

Finally, when I meant frame rate support what I meant that USB/HDMI device needs to have a proper HDMI processor that supports a reasonable amount of video modes that most cameras output.  The actual tablet device display frequency is independent of the HDMI feed.  In terms of data plumbing, if your tablet can reasonably display YouTube videos using H.264 then the Inogeni device should work because UVC 1.5 protocol at its best resolution supports H.264.   Again this assume you can shovel the HDMI bits into the tablet correctly and fast enough.

The sad thing is that an 8" to 10" tablet with HDMI input will never be built.  Its just too esoteric/ specialized.  Large companies who make commodity devices with razor thin margins cannot afford adding such features.  Look at the demise of Colby.  However, a small company with a niche product like the Inogeni can gouge video enthusiasts with a $400 device.

On 2/12/2016 at 2:36 PM, blafarm said:

I wouldn't count on this being a usable solution.  Here are the potential problems I see in no particular order:

1.  Android device must support USB 3

Yep.  Only a few tablets support USB 3, such as the new Asus  ZenPad S 8.  I imagine for 1K capture USB 2.0 should be sufficient.  One can buy a ton of HD webcams on Amazon that use only USB 2.0.

On 2/12/2016 at 2:36 PM, blafarm said:

3.  Assuming the requirements specified in Number 2 are satisfied, the OTG cable must support this capture device

The OTG cable probably just needs to support USB: its like saying an Ethernet cable must support Skype.

On 2/12/2016 at 2:36 PM, blafarm said:

5.  Assuming the requirements specified in Number 4 are satisfied, the solution must not require the Android device being rooted (for obvious reasons)

This device is dependent on Web Cam (aka UVC)  support built into Android so there definitely no obvious need to root the device.

On 2/12/2016 at 2:36 PM, blafarm said:

6.  Assuming the requirements specified in Number 5 are satisfied, un-rooted Android devices currently cannot simultaneously function in USB Host Mode and be charged at the same time. Currently, you get to choose only one -- an Android device is either a USB device, a USB Mass Storage device, or a USB Host device.

The Android tablet would just run in Host mode.  People have been able to plug in webcams to Android tablets for a while now.

On 2/12/2016 at 2:36 PM, blafarm said:

7.  The latency of the entire signal chain (capture device, data transmission, device-based app processing, device-based display output) will likely be unacceptable for real time applications.  Of course it's impossible no know, but I would 'blue sky' estimate the latency as being somewhere in the ballpark of the Sony A7x PlayMemories Android app.

I've actually researched the feasibility of this setup a few weeks ago.  There is no reason why a wired system wouldn't work reasonable well.  The issue with WiFi based systems is not so much latency, which does matter, but the time outs and drops due to poor connectivity, radio interference and what not.  Basically, this device is fancy web cam except that the built in camera is replaced by a HDMI cable input.  There should be zero connectivity/drop out issues.  Latency should be minimal.

Companies actually sell silicon chips that are called web cam processors.  These chips handle all the USB interfacing to the "computer."  All the board designer needs to do is wire in a video camera.  All webcams have this chip built inside.  This particular adapter is just reading the data stream from the HDMI connector, writing this data into a memory/frame buffer and then shoveling that into the web cam processor.  Pretty simple actually. Not $400 simple, more like $100 simple.

On 2/12/2016 at 2:36 PM, blafarm said:

There is a reason no one is doing this -- it's just not practical.  Generally speaking, Android devices are architected as media consumption devices -- not content creation.  Over the years, I have run into a couple of Canon-based apps that display Live View with minimal latency.  And, there are small number of Android devices that have HDMI inputs -- but their form factor would not satisfy this need.  As much as I would like a solution like this, I've concluded that a dedicated monitor is the only viable approach.

 

Practicality is quite subjective.  I think many people once thought shooting video on a DSLR was neither viable nor practical.  Also, Canon Live View is proof that this idea is feasible because its basically doing the same thing. Also, I am not aware of single Android device that HDMI input.   People already have tablets and huge phones.  It would seem desirable to at least non-professionals to be able to reuse a nice 8" tablet for for framing if they already own the device and are probably carrying it anyway.  That said, $400 seems crazy expensive.

Storing video is another issue. Most tablets don't have SD slots.  Also, micro SD media capacities aren't so great nor can you generally easily swap out a SD card on an Android device easily without rebooting the system.  More of an annoyance.  

I would be concerned with really mundane stuff like does it handle 24p, 30p, 60p frame rates and 3:2 pull down properly.  Even a "big" company like Black Magic is having issues with its Video Assist with HDMI out from Sony cameras.  

More importantly, is how is this video being recorded.  The latest UVC 1.5 standard supports H.264 at best.  Since Inogeni isn't supplying the software, what software does one use to record the video?  Is it any good?  Does it support a H.264 stream?  

On 2/12/2016 at 7:26 AM, Luca Acerbi said:

Is there any technical reason for the 8bit limitation?

Probably a limitation of UVC protocol built into Android.