December 20, 2014, 12:24:13 PM

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - peederj

Pages: 1 ... 3 4 [5] 6 7 ... 20
61
HDMI is a data transfer protocol standard, just like TCP/IP or USB. And it's a dumb consumer-level standard...its professional sibling is called SDI. And on the C300 you get both options, and Atomos makes a higher end Ninja they call the Samurai that just does SDI instead of HDMI. HDMI was for HDTV, which for legacy reasons uses 1080i60, sending half the lines 30 times per second, alternating with the other half (all interlaced together) also 30 times per second. There is nothing a camera or anything else can do to alter the protocol...that's how the machines have all agreed to talk, and because it's so dumb a standard, you can expect any HDMI monitor or whatever to understand it.

But we don't want to shoot 30p or 60i (both are offered on the C100 btw), for cinema the standard frame rate is 24p, which gives a nice familiar texture and blur to motion when captured at a shutter angle of 180 degrees (1/48th of a second or 1/50th on the 5D3 is close enough). So how does one send a 24fps signal over a 60i protocol? I forget the technical term (you can google all of this of course, and please do) but they essentially just repeat frames alternating 3 repeats and 2 repeats to match the timing difference of 30:24. And the receiver, to deinterlace that redundant 1080i60 stream to 24p (23.98 is the actual rate on the Ninja also due to legacy concerns with synchronization) must figure out the cadence of repeats and drop the appropriate redundancies, leaving a steady and even stream of progressive frames.

The Ninja has to do this empirically off a moving image being sent so it can see which images are repeats and which are new and drop the proper ones. HDMI can't communicate that information sadly, meaning the Ninja waits with record disabled until it gets enough contrasty motion to see what to do.

As for your growth plans, I understand dipping your toe in, but a lot of this only makes sense when you have the full professional rig in front of you and understood, and then you can learn to make do with less. Good luck with your exploration however you go about it.

62
The Atomos Ninja 2 costs $1000 and comes with everything you need, and also provides a very useful 2nd monitor with peaking, false color etc. It's small and the batteries and media are cheap as chips. Six hours of recording ProRes HQ onto a bog standard 500GB 2.5" laptop drive. SSD optional. Pressing record on the C100 starts and stops recording automatically on the Ninja. Your laptop is going to be a complete pain in comparison, plus you will need some form of HDMI capture ability for it.

The only minor annoyance with the Ninja 2 is it needs to analyze a moving image to properly deinterlace to 23.98fps 1080p. The HDMI standard is 1080i60, and the C100 (and 5D3) are going to have to send that format out regardless. You can record that directly but deinterlacing in post is a pain. So instead with the Ninja 2, you can have it deinterlace and record 23.98 fps directly, which is wonderful, but every time you power both the camera and the recorder on, you will have to wave your hand in front of the lens a bit for it to figure out how to do the 3:2 pulldown.

A nuisance but that is the only nuisance...everything else is beautifully optimized and integrated. You can name the project, camera, scene and shot directly on the Ninja and the take number will auto-increment. Timecode is fully supported. It's a professional solution. The Hyperdeck Shuttle 2 is actually bigger and far less featured, and while it's half the price (with the needed mount plate) it mandates the use of SSDs which make up for the difference. Also, for the 5D3, they may continue to insist on turning off the internal LCD when using HDMI, so with the Hyperdeck you won't be able to see what you're shooting. Get the Ninja.

But if you don't care at all about ergonomics and workflow, why are you caring so much about image quality? Do you just shoot product shots or something in one studio?

63
On the topic I can confirm that the C100's HDMI output is clean uncompressed 8 bit 422, with timecode and record start/stop control. So with the Ninja 2, the ProRes 422 HQ 220Mbit is a very high quality image to my eyes, the noise floor does show the one dimension of chroma subsampling (444 would be better) but is fine and very even without much banding and no macroblocking.

The C100's compressed internal video is fairly ugly in comparison, I'm just using the SDXCs as safeties and doing all recording onto the Ninja. I strictly use Canon Log gamma to get the 12 stops of dynamic range this way, I don't notice banding in color gradients as 12 stops over 8 bits is still very high fidelity even though 10 bit would be optimal for 12 stops. ISO 20000 is quite usable...I think I will use the NDs instead of going below the native ISO of 850, but going below that does crush out the noise floor at the expense of recoverable dynamic range. I'd rather use NR in post to taste.

I would say the C100 is at least a stop better low light than the 5D3 video, though I will want to see what the Ninja does with the 5D3 clean HDMI output, which Canon said would be uncompressed 8 bit 422. Noise is a very complex thing for a codec to handle of course, and chroma subsampling makes color edges the other main concern. I have great hopes for the 5D3's clean HDMI out, it will still not be as comfortable a video camera as the C100, nor will the downsampling be as sharp. But it will have the full frame look and DOF and be free of moire and aliasing, and the noise floor should be quite fine and color edges resolved well. The dynamic range might be OK with Cinestyle gamma in that case, right now I don't use that because of the codec. Canon Log is still going to be better I predict...would be nice if Canon released Canon Log for the 5D3 along with the clean HDMI.

I'm sure Canon delayed the 5D3 clean HDMI till April just so people interested in doing films would give the C100 a try...I think the C100 will be my A cam workhorse for years even if the 5D3 video lives up to my greatest hopes. But I do want the good B cam, and I don't mind picking up a second Ninja if in April things work as I hope. Enjoy your shooting, Canon is being less and less evil these days, and the C100 + Ninja + Canon Log is a dream come true for me anyway.

64
EOS Bodies - For Video / Re: 5dm3 picture style for video
« on: November 02, 2012, 09:53:27 AM »
I shot the typical Neutral 0, -4, -2, 0 for a year but really Neutral is quite brown and dingy compared to Faithful that has the same dynamic range benefits but a warmer and richer skintone. I also don't think -4 contrast is so good an idea, so I currently shoot Faithful 0, -3, -2, 0 and then add a bit of sharpening in post, a bit of desaturation usually, and a bit of contrast if needed.

The important thing is to get as much as you can through the codec while not requiring tons of manipulations in post. I tried Cinestyle but that's not so good an idea, you only have 8 bits to work with, and 4:2:0 until clean HDMI appears. The most important things are white balance and exposure...I set custom or kelvin white balances religously. To minimize burned in noise, I use the pull-down ISOs a third stop below the standard ISO (e.g. 5000 instead of 6400) and NEVER use pull-up ISOs (a third stop above standard). I never use HTP but some noise reduction is up for debate.

65
Lenses / Re: Canon EF 40mm f/2.8 Pancake
« on: October 22, 2012, 02:23:00 PM »
I'm doing the standard thing of using it as a body cap so I can grab quick impromptu shots as needed. It's not truly great but it's better than missing the shot or burdening a group with preparation. I wouldn't rely on it as a primary but it's handy.

66
Lenses / Re: A very dumb view
« on: October 19, 2012, 09:18:09 PM »
Is there a site that scientifically tests and ranks filters? I would enjoy such a site. Certainly the anecdotal reviews we get from people trying to justify their spending are unlikely to be reliable.

67
Let's discuss the heat theory.

First of all, the 1DX already shoots video at the same frame rates of the 1DC. The 1DX also does full resolution stills at roughly half the speed of that video. And as Canon's flagship stills camera it's unlikely they crippled the 1DX still image any with Nikon breathing down their back.

If Canon is doing additional things to cool the sensor for 4K video, to reduce noise and improve low-light performance, why wouldn't they do that also for the 1DX? I doubt it is sensor cooling.

The 1DX video does not line skip but instead reads the entire sensor at runs the result through a downsampling algorithm. Unlike the 5D3 it does not do pixel binning or any other shortcut. So the 1DX, sensor-wise is going to be the same in its operating heat as the 1DC.

The 1DC otoh doesn't downsample the sensor readout but instead crops it for 4K resolution. This is the computationally easiest operation resolution-wise, as easy as line-skipping was on the 5D2. Just discard information out of the frame.

It's conceivable that Canon improved rolling shutter performance for the 1DC over the 1DX by increasing readout speed. They haven't mentioned this, but if so, it will be very easy to test (mount both cameras on the same fluid head, pan back and forth and compare the distortion angles of vertical lines). Canon has the tech in-house as the rolling shutter performance of the C300 is excellent.

If that is not the case, then the only other matter is whether the DIGIC chips and CF cards will get significantly hotter processing 4K at 30fps than they do processing 18MP RAW+JPEG at 15fps (forgive and correct me if I don't have these specs memorized right). It's conceivable but I don't think it's so very likely.

They haven't added a fan to the body, they may have tweaked the heat sink approach a bit on the processor chips, but I have doubts that they could do much to improve a fundamentally passive design without visible modifications (such as metal fins or something).

So I call BS to the heat theory. Where is this additional heat coming from, and more importantly, where's it going?

It's just firmware dude. They may put something in there to make you think otherwise but there are so few of these being made at that price that I doubt it will be much.

68
EOS Bodies - For Stills / Re: Canon EOS-1D X DXOMark Sensor Scores
« on: October 14, 2012, 04:30:47 PM »
The conspiracy theorists would suggest DxO is waiting the completion of "sponsorship negotiations" as to what the final score will be.  :o

69
EOS Bodies - For Stills / Re: Canon EOS-1D X DXOMark Sensor Scores
« on: October 13, 2012, 07:51:31 PM »
Mikael,

FWIW, you're even more nuts.

Hugs and Kisses...  :-*

70
EOS Bodies - For Stills / Re: The Unholy Trinity of Non-L Primes?
« on: October 13, 2012, 07:50:16 PM »
The 50/1.4 and 100/2.8 Macro are my picks for non-L primes in the Canon lineup. I ended up liking the 40 shorty a lot and have it as my body cap and shoot more with it than I should. On the wide end, potentially the new 24 or 28 IS versions are good, but I haven't tried them. I have the Zeiss 25/2 in that range, which makes sense as I do a lot of video. With stills I am usually using zooms (Canon 16-35 II, 10-22) on the wide end, fast apertures aren't so very crucial there but proper framing is (want to maximize resolution).

71
EOS Bodies - For Stills / Re: 1dx, or d800e?
« on: October 13, 2012, 07:42:22 PM »
Rent the D800e for a weekend and see if it's all that.

If I'm you and money is this much of an object, I'd get the 200/2, maybe selling off the 7D and picking up a used T2i as a backup if you feel you need one.

Having to juggle two completely different systems and mounts would drive me nuts personally. If you're leaving Canon, leave completely.

The D800e has a better sensor and Nikon has the 14-24 and that is certainly a nice landscape kit. You don't have the money it sounds to play in that tool-for-the-job field, so stick with what you have and get more wedding$ till you do.

72
EOS Bodies - For Stills / Re: Canon EOS-1D X DXOMark Sensor Scores
« on: October 13, 2012, 03:36:12 PM »
I've come to the conclusion that the 1DX is worth paying double for vs. the 5D3. With Big Megapixel Talk on the horizon though, and Canon needing an answer to the D800e (which has a damn fine sensor and that's about it), maybe it's better I wait a bit before trading in this camera that I only got this year.

73
If Photographer B intentionally lobotomized Photographer A, ruining their photographic and earning ability for the remainder of their useful life, Photographer B would at minimum be immoral and unethical.

OK, we're getting a bit beyond the pale, but there are valid arguments that this behavior is immoral and unethical in that there exists some degree of compact between consumer and manufacturer that one won't unfairly take advantage of the other. The fact of transition costs from system to system (lenses etc.) mean that this can indeed rise to the level of betrayal. You can say that we should have known there was such a possibility beforehand, but that's like saying Photographer A should have known Photographer B might have been a sociopath.

It is possible that this has veered so far off-topic that there is no getting it back on, but...

It occurred to me it's ironic and hypocritical for photographers to even be having this argument.

Photographer A and Photographer B both shoot a wedding. Photographer A charges $800 and Photographer B charges $8,000.

Photographer A's images are out of focus, improperly exposed, mundane and uninspiring. Photographer B's pictures are not only technically perfect, but they are absolutely luminous, capture the moments perfectly and positively soar.

But wait, Photographer B admits that he USED THE EXACT SAME CAMERA as Photographer A. All of the difference is in the software holding the camera.

Clearly Photographer B is immoral and unethical.

74
EOS Bodies - For Video / Re: How canon charges 6000$ for firmware upgrade
« on: September 21, 2012, 04:34:46 PM »
Any software development types here that can give the masses an idea of how different the code would have to be to handle the 4K video stream?  It seems like the general thought is that its easy/free to 'turn on' this feature, but I'm guessing there is more to it than that.

-Brian

It may in fact involve doing _less_ rather than more in the pipeline. If they use the native resolution of the sensor for their 4K (which they do, it's simply cropped to native res) you won't have to do a downsampling step. The rest of your processing can be the same pipeline as stills as long as the processors can handle the throughput without power or heat problems. The codec at the end will have more data to process, but the codecs can handle that and are industry standards. In both cases, the 1DC proves that the stock 1DX hardware is capable of the feat.

If that is the case, then the 1DX is exactly a _crippled_ version of the 1DC: resolution-destroying code is inserted into the firmware of the 1DX that is left out of the 1DC. This might not be terribly hard to hack; I can think of two approaches off hand instantly.

75
Software & Accessories / Re: Aperture 3.4 Issues
« on: September 21, 2012, 12:05:07 AM »
Thanks.

Roll over, Steve!   :-[

Pages: 1 ... 3 4 [5] 6 7 ... 20