I have to give my subjective opinion in this debate. Looking at your night test EOSHD i've to say that the 'look' of the mjpeg is exactly what i'm searching for. I find it really film looking. I notice and i recognize that the avchd codec gives more details, less noise, less banding, but i see to much microcontrast, blacks are too black, edges are to... edgy, i don't know how to explain. What i really wish is a codec, probably the avchd, that has also the qualities of the mjpeg. I hope that the research on quantization can lead to these results.

ughhh, you still don't get it andrew and you say everyone else is blind . I fully understand why you wouldn't want to spend your time on mjpeg lowlight and I'm actually with you, I don't care for mjpeg really and I can achieve it's look in post. But it's plain insulting to the intelligence of other people when you say it's a "total brand exercise and useless". You say: " I don't like to see people waste their time on something that's so obviously as flawed a concept as 'an MJPEG low light patch'." I'm sorry, but who promoted you to our time rescuer hero? I and only I will decide on my time distribution, and frankly, since I make a living on video production etc, I think I'm qualified to make an artistic decision that might involve not choosing avchd in lowlight (" If you shoot in low light at high ISOs on the GH2 stick to AVCHD, period"). About the whole business concerning your EOSHD anamorphic guidebook, I'll tell you why it came up. A brand new thread on the GH2 hack which gives you an extra option, worthless or not, and you attack it calling it a total brand exercise, now that's personal, and so there were personal answers to that about your own, rather epitomized, "brand exercise". Simple really.

I'll talk no more about this, since hopefully I explained some stuff for you

to end this I'll quote your own words: " there's enough room in a world of 6 billion for different philosophies and approaches to one's perspective on life, and each person has more than one!"
enough said!

Stefanos

@stefanos
Please continue to debate with me but do address the issue, not me. It isn't a personal crusade it is a suggestion, a firm opinion. The issue is the tweaking of E1, E2, table, quality, and calling it 'low light patch'. It is not a new patch, it is the same as the last one but with a different label. It doesn't give us anything new, it doesn't lead to anything new. You can put E1 quality to 400, 500, 600, 700 it doesn't change the fact that AVCHD is better in low light, it just encourages people to waste time and to shoot with the wrong settings in low light. Those with an artistic need for 30p and more noise - fine, no problem with that at all. But it is not really the point here. The goal is to improve the low light performance of the GH2 and also if possible do research that leads to better MJPEG. Just huffling bitrate again and again doesn't do that. I see through my site that the multitude of MJPEG quality / bitrate patches are simply a distraction for people who want to use the hack in their projects as early as possible. I also reckon that it gains so much attention it's a distraction from more meaningful testing. How about a brute force coordinated approach to AVCHD low / high numbers in the latest PTools instead? That has the potential for better low light if I am not mistaken. I am not saying everyone should suddenly stop MJPEG research all I am doing is stating an opinion. Feel free to ignore it if you want, but be mature enough not to suddenly think I am some kind of kitten murderer or something. You should define people by meeting them not by their opinions on a forum.

@Vitaliy_Kiselev
fair enough, I finished what I had to say with the comment I just posted, why was it deleted so easily? It wasn't inappropriate or anything, the opposite it was somewhat constructive. Unfortunately, Powel's topic became flame and thus it was moved here, so that people's views wouldn't be misrepresented.... Anyway

topic title is "EOSHD crusade... on MJPEG hacks" :) if it was named "MJPEG HACKS" we wouldn't even be discussing this right now

sorry, but I love both modes in low light,
if I want a little more "Filmgrain" I use MJPEG,
what's wrong?

@Butt: exactly.

I am not on a crusade against MJPEG per-see it's that I just don't think 99 combinations of bitrate tweaks to the same settings that have been around since the GH1 are going to get us very far.

Some people think image quality and bitrate are exponentially linked somehow. They're not. There is no point pushing beyond 100Mbps when you still have 8bit, 4-2-0, 720p upscaled with moire, and just 30p. In fact you'd really need a magnifying glass to see the difference between MJPEG 30Mbit and 100Mbit as it is. I just think it's the wrong thing to focus effort on, that's all, nothing personal.

How about some research into the 4MP burst mode at 40fps and seeing if the MJPEG codec can be used to capture a feed with the sensor mode in 4MP stills burst mode... Now that would be a step forward.

>How about some research into the 4MP burst mode at 40fps zed seeing if the MJPEG codec can be used to capture a feed with the sensor mode in 4MP stills burst mode... Now that would be a step forward.

I would also be interested to know if Vitaliy has been able to see anything about that mode in his low-level GH2 exploration (not for more 4MP pictures burst but for less, better quality silent shutter stills)

Also, your continuous 720p reference shows you haven't been researching this much. You're right, on the GH1, 1080 MJPEG was just an uprezzed 720. On the GH2 it's not, it's better than that.

Do YOUR research on this subject and then bring it to the table.

Well put stonebat. Please keep testing!

As for the claim that patched MJPEG frame sizes (on either GH1 or GH2) are literally resampled from a 1280x720 frame, I still maintain that this is an unproven speculation.

While MJPEG 1080p videos tend to look less sharp than AVCHD 1080p, that does not prove that they were in fact upscaled from a lower resolution image. Since Panasonic has chosen not to reveal the algorithms it uses to produce video frames from the high-resolution Bayer Filter raw data output of the image sensor, we may never know for certain how it's actually done. However, if a second stage of upscaling was inserted after a fixed-size 1280x720 frame was generated, that should produce telltale periodic irregularities in the width of finely-detailed diagonal lines in a test image. To date, I have seen no tangible evidence of any type of rescaling artifact of this sort in any MJPEG video.

In all G-series cameras, the MJPEG encoder is required to generate not just a single HD 1280x720 frame, but multiple WVGA, VGA, and QVGA resolutions and aspect ratios as well. In addition, the MJPEG encoder is evidently flexible enough to work with a variety of image sensor scanning modes and sizes, ranging from 12-16 megapixels. Given these diverse requirements, I think it more likely that the encoder uses a single general-purpose scaling module capable of being programmed (or patched) to produce any specified frame size from the scanned image sensor aspect ratio. In any case, that is effectively how the MJPEG width and height patches work in practice.

I hear you @LPowell

At least 480p MJPEG seems much improved with the latest PTools. Before (with previous release) it looked like VGA mode simply upscaled. Now it looks like the actual scaling resolution 1:1. What has been changed to enable this Vitaliy?