Thursday, January 10, 2013

Smart cellphones as audio recorders - OK casual - Pro needs better

Continuing story .. smart cellphones as audio recorders.  We can download apps for most of them which record wav files.  Exciting idea - can this give us readily available high quality indie audio recorders?  I now think "no" as in not yet, or not with the phones we/friends/family currently have.

Story so far: Good early results with voice recording for film with the very low cost "Vodafone 858" - a Huawai Android.  More use as a backup recorder reveals electrical interference issues eg nearby electric fences.  Ref earlier post "Do Androids Dream of Electric Fences"  here:
http://iafilm.blogspot.co.nz/2012/12/do-androids-dream-of-electric-fences.html

New results:
I also find that high sound levels in music performances cause the little Android to distort, as in the app or the phone does not do automatic level setting or needs to do it better.

This last week I have been getting to know a Windows Phone 7.5 LG Quantum from the older lower end of the current Windows Phone system.  Similar result - OK for informal tests and student exercises but not quite good enough for serious indie production.  No result yet on electrical interference.  Loud sounds cause distortion like I observe with the Androids.

Most interesting Windows Phone result is that with a variety of recording software including an "Ultimate" app I paid for, the sample rate was always a relatively low 16khz with no controls for increasing that.  I am a computer programmer, mostly "Windows", so I set out to study microphone programming.  Interesting finding with wider implications is that the microphone behaviour is fixed and programming starts with the chips passing the software audio that is already digitised at 16khz.  So even with original programming with the "Dot Net" system I can do nothing about that 16khz.  The "xna" programming library of functions also offers no control over recording level.  I am intrigued though to discover in "OneNote Mobile" a notes sound recording function which gives very low quality but does seem to give an automatic recording level effect as in I speak very loudly into it and it copes.
QUESTION - Is microphone behaviour locked into the silicon chip design, or is there some deep low-level programming in the wider "Windows" outside of "Dot Net" that can change this?
QUESTION - If microphone behaviour is in the chip, would different brands of phones with different chips behave differently?  If you are reading this and you have a smartphone, you can use the blog messaging system to contact me and join in the test recording fun going on here.
THOUGHT - This experience suggests that Phone programming is about glueing together sub systems that are very much already programmed for us, and the internal-based parts of  "Apps" are really about alternative presentations of what is already there.  ie "Apps" are relatively easy and also limited - until we get into the area of doing more by connecting over the internet to a high-powered server to add more power to our problem-solving.


Tuesday, January 8, 2013

Camera upgrade - selected Sony NEX 5N - results Part 1

Some past posts were about what camera we should use for our upcoming green screen experimental feature.  Made selection, done deal, I went for a Sony NEX 5N
This was on a special reduced price here in New Zealand.
I am guessing that had something to do with newer NEX models coming out.
The reviews suggest that this model 5N does video better than the newer ones.
I had some concerns at the need to buy a 50fps  PAL model.  I did consider trying to buy a secondhand 60fps NTSC example from EBAY.  Some web searching indicated that most target film festivals can handle 50fps or the obvious 25fps derivative video.  NTSC conversion is do-able for us based on experience with our short film "Laputa" as described elsewhere in this blog.  
[Aside note - I suggest that this whole 2 alternative model thing that camera manufacturers do has become obsolete and they should sell us NTSC everywhere for consumer and semi-pro cameras.  Most computers systems are effectively NTSC and all modern TVs and Projectors can do both.  The flourescent light flicker rate issue is easily fixed with shutter speeds of 100, 50 or 25.]

Some first results and impressions.

Green Screen
YES - best results I have ever seen from our testing, which is a relief after buying this camera based on reports from others.  Test here is where the subject (me) is moving around a lot - story is riding in an SUV on a bumpy road - this is the test that has caused the most problems for other tested cameras.  This image is taken from the video in the middle of one of my highest speed action movements - leaning sideways.

Tech.  Kit lens manual settings with slight under-exposure. Shutter speed 1/100 sec.  Tungsten lighting.  Scene setting "Normal".  Video best setting 50fps and 28Mbps.

Moire problems?
NOT FOR US.  Some reviewers state that there are "moire" issues with the NEX 5N but we have seen nothing in 3 weeks so far of operation with a variety of tests and 1 production shoot so far.  But note that the way we film puts us at low risk of moire issues.

Use prime lens collection?
YES - this works well!  I have now bought a collection of lens adapters from EBAY store "jiakgong". 1980 vintage lenses from my Russian Zenit camera adapt just fine and working with them is straightforward.  I am about to start testing borrowed Nikon lenses.  Images following are an experiment in taking the zone-focussing capability to artistic extremes.

1. Kit Lens, moderate zone focussing ...

2. Russian Helios Lens f=58mm, aperture=f/8, shutter=1/100

3. Russian Helios Lens f=58mm, aperture=f/2, shutter=1/500 - ZONE TO THE MAX


Can NEX 5N make my 1953 vintage Bolex 16mm movie camera live again?
PARTLY - It is possible to adapt the 1953 C-Mount lenses to the NEX 5N but they cover an area just over 10mm wide in the centre of the 20mm wide sensor.  This area is about 2000 pixels across so in theory with a re-sampling digital zoom control the old 16mm lenses could ride again.  BUT the NEX-5N digital zoom switches off for video.  Therefore this lens collection can deliver reasonable stills, but for video the result needs to crop to about 800 x 450 pixels.  ie NEX 5N can party like it's 1953 but only in standard definition.  I read that new NEX models coming out starting with the NEX-EA50 will have a digital zoom that can resample.  This leads me to a cheeky comment about the "Digital Bolex" project to make a camera that is the "spiritual successor to the Bolex".  My opinion: guys you are too late - NEX has already got there and it looks like Panasonic is heading in this direction as well.