Quantum X 'SpanMyBitchUp' v2b vs. 100Mbps Flow Motion v1.11

Marco

What are the advantages / disadvantages between "Driftwood Quantum X 'SpanMyBitchUp' v2b" and "100Mbps Flow Motion v1.11", as both of them are roughly 100Mbps?

Which one is better for general documentary work (24p and 25p)?

Which one delivers better video quality at fast moving objects?

driftwood

Test both out and whatever YOU prefer, choose.

tinbeo

@driftwood: I was thorn between yours and Lpowell setting. Using avid and mts neeeded to be transcode and it took forever. I am going in a run and gun situation. shoot and edit and upload to server of a news TV program. I like to use your. But which setting you would recommend for Mjpeg which Avid could see and edit without trans code?

driftwood

@tinbeo My MJPEG is the highest quality setting yet on GH2 from what Ive seen :-)

mrbill

You just import the files directly into avid at the resolution of your choosing - .mts files are a lot quicker than the .movs that the canons generate, whatever patch you choose to go with

paglez

Are there any reason to use MJPEG patch instead of AVCHD patches? MJPEG has a visible croma noise and a very serious time limit also. Why use MJPEG?

tinbeo

mjpeg work with ama link in avid. That 's just for me, I think, for a run and gun, shoot and cut no need to transcode

LPowell

@paglez In low light situations where you must underexpose by more than one stop, Flow Motion's MJPEG patch will maintain higher bitrates than AVCHD, with image quality comparable to Photoshop's Maximum JPEG quality.

paglez

@Lpowell ok, but this is a very concrete scenario. Most of situations AVCDHD is clearly superior

LPowell

@paglez Yes, except in low light.

kavadni

@LPowell "In low light situations where you must underexpose by more than one stop, Flow Motion's MJPEG patch will maintain higher bitrates than AVCHD, with image quality comparable to Photoshop's Maximum JPEG quality."

I may have missed a point completely somewhere. Is MJPEG going to be better than a/the AVCHD CBR patches?

tinbeo

@LPowell if you could make mjpeg setting span like Canon XF 300 then GH2 should be super

LPowell

@tinbeo But then you'd have only two more wishes left. Wouldn't you rather have a pony?

rockroadpix

lol

Marco

I now took quite some time to compare this two settings:

"Driftwood Quantum X 'SpanMyBitchUp' v2b"
"LPowell 100Mbps Flow Motion v1.11"

Unfortunately the BMP files are too big to upload them in the forum. Therefore you can download them here. So everybody can compare the screenshots for themselves. My internet connection is not fast enough to be able to upload the original MTS-files:

valicelly.com/middle-flow-24P.bmp

valicelly.com/middle-flow-25P.bmp

valicelly.com/middle-span-24P.bmp

valicelly.com/middle-span-25P.bmp

valicelly.com/right-flow-24P.bmp

valicelly.com/right-flow-25P.bmp

valicelly.com/right-span-24P.bmp

valicelly.com/right-span-25P.bmp

MTS Average MBit/s:

"Driftwood Quantum X 'SpanMyBitchUp' v2b" @ 24P: Average 76 MBit/s
"Driftwood Quantum X 'SpanMyBitchUp' v2b" @ 25P: Average 62 MBit/s
"LPowell 100Mbps Flow Motion v1.11" @ 24P: Average 28 MBit/s
"LPowell 100Mbps Flow Motion v1.11" @ 25P: Average 17 MBit/s

All tests have done with the GH2 and Lumix 20mm lens. Unfortunately I did not had a manual focus lens. Although I had set my camera on manual focus, after updating the firmware, it also reset the focus, Therefore I had the refocus again and therefore the focus when shooting the two different settings is not exactly the same. Therefore the shots are only limitedly comparable.

However, these shots are very interesting and useful to compare the quality difference of this two settings concerning 24P and 25P.

In order to easily be able to compare the screenshots you need a Full HD display and you should look at the screenshots in full screen mode (for example with IrfanView, which lets you to navigate through the photos in full screen mode with the left/right-arrow keys of the keyboard). The screenshots have been exported with Premiere Pro in BMP, so there has not been any compression of the screenshots. I've opened the files in Photoshop and inserted the titles and then saved them as uncompressed BMPs again, so there hasn't been any compression degredation.

My findings concerning quality difference between 24P and 25P:

"Driftwood Quantum X 'SpanMyBitchUp' v2b" - 24P vs. 25P:

The resolution between 24P and 25P is the same as good.
Compression artefacts are a little bit more visible at 25P. Compression is not visible at 24P.
Diagonal lines are a little bit more jagged (aliasing) in 25P than in 24P. This is the camera's fault.

"LPowell 100Mbps Flow Motion v1.11" 24P vs. 25P:

Compression at 25P is so strong that 25P footage is not usable for pro use.
Resolution at 25P is much lower than at 24P.

Because of the focus difference in my test shots (see above), it is difficult to compare SpanMyBitchUp with Flow Motion in my screenshots. However, in my setup (my simple moving test chart), it can be said that at 24P the difference in resolution and compression artefacts between SpanMyBitchUp and Flow Motion is very subtle.

However, at 25P I strongly recommend to use SpanMyBitchUp instead of Flow Motion. I hope there will be an update soon which will improve Flow Motion at 25P.

Kihlian

@Marco Thx for testing. This was my feeling as I tried Flow Motion at 25p.

mrbill

Interesting. I've just completed a day's interviews using Flowmotion HBR 25p, and the IQ looks great. I've had no urge to pixel peep at all.

LPowell

@Marco

"LPowell 100Mbps Flow Motion v1.11" @ 24P: Average 28 MBit/s"

"LPowell 100Mbps Flow Motion v1.11" @ 25P: Average 17 MBit/s"

While I can't be sure what happened to cause such low bitrates, this indicates severe bitrate starvation. Under those conditions, I'm not surprised you found the 25p image quality lacking in detail. Flow Motion v1.11 should normally produce bitrates from about 35 - 100Mbps.

Another factor you may be able to confirm is whether the sample BMP's were captured from an I-frame at least three GOP's from the start of the video stream? A well-known characteristic of the GH2's AVCHD encoder is that the first couple of start-up GOP's are typically of lower quality than the rest of the stream.

Marco

@LPowell

Yes, I also was wondering why the bitrates where so low. I hope I didn't make a mistake. Probably my moving test chart is not good enough. When the next version of Flow Motion is out, I will make another test.

By the way, I have to correct my statement regarding Flow Motion that "Resolution at 25P is lower than at 24P". Actually it's only the compression artefacts which are much more visible.

I made all screenshots more than 10 seconds after the start if the video stream.

DrDave

I don't get those low bitrates. Even in an average indoor scene with one 60 watt light bulb I get 30-40 on Flowmotion's lowest setting. I can maybe get 28 if I shoot an image that has no detail in it. As for pro use, well, pro's use it. I'm sure pro's probably don't use it at a bitrate of 17.

With Flowmotion in interior, less detailed scenes, the beauty of the system is you simply bump up the setting if you are getting low bitrates, and still get stable performance. Recommend you adjust the video until you get the right settings, at least around 45, if you are comparing it to a codec at 60 or so, and check you have reasonable bit rates before doing the comparison. Otherwise, why bother? My four year old camcorder, which I only keep because it has a firewire port for streaming, has a bitrate of 18.

Looking at the sample, you have several flat, dull angled surfaces. These are good for testing noise, but they won't stress the codec. Also, to my eye the details do not look as sharp as I would expect from a GH2 with a prime lens. To test a codec I normally use F4 or higher on a prime lens, and I use 5.6 if I want to test spanning. Not to say I can't shoot for shallow DOF, but if only a fraction of the image is in focus it isn't going to tell me much about how the camera is capturing data. The Flowmotion looks about a half stop underexposed compared to the Driftwood. Anyway, it isn't as bright. Exposure has to be the same as it makes a big difference in the results.

I like the idea of static objects with the small fountain. Hope to see a test without the low bitrates. After all, no one would use the hack with low bitrates. You can also tack a chart and some brightly colored objects on the back wall. I found in my own testing the different patches had enormous differences in the way they render the bokeh objects, and that this in turn had a big impact on the overall quality of the image. In other words, the objects IN focus look more or lessthe same--only PP differences, which are not really relevant--the objects out of focus vary substantially.

Not to say the driftwood doesn't look good, because it does look good.

DrDave

And thanks for doing the test

Rambo

If Flowmotion is looking that good at 17-28MB/s, imagine how good it is at 40-80MB/s. I've not seen bitrates that low on anything I've shot using Flowmotion either, something's wrong.

ghostlayer

I shot nice footage using flow motion patch, if think the actual bitrate was close to 80-90 mbs (using SD class 10 - 30 Mo/sec) I found HBR 25p suitable to Pro shooting. I would be very intersted if others pro users could tell me if smack my bitch is even better ? Does smack by bitch up allow direct in-cam playback? And what about aqua motion ? Thanks for sharing

lunalobo75

@ghostlayer SMBU v2b is a great patch/setting. No card errors for me on any modes (3o Mb/s Sandisk Class 10 HD extreme), all modes played in-camera for me, and the video is stunning. It's a patch no one talk about much.

ghostlayer

Thank you lunalobo75 ! I think I'll try it ! I like underrated patches :) I guess ISO can be unlocked up to 12 800 (at least in ptool end user setting)

peternap

@ghostlayer Both are great patches and it's really a coin toss, but I have had a couple of card errors using 45mb Sandisk cards with SMBU, but can't force one with FM. Card errors are unacceptable for run & gun IMO.

Howdy, Stranger!

Categories

Tags in Topic

Top Posters