Sony Movie Studio 12 – CPU versus CUDA rendering test …

So I have a new role at work … (wow, non-sequitur much?)

Ok, well, that was a rough start. What I meant to say is that I have a new role at work. This requires that I work with some of our larger software, and so I have built a large VMWare image with a 10GB RAM profile, stealing most of the RAM in my 16GB RAM machine. So of course I upgraded half of it to 8GB sticks giving me 24GB in total and some much needed relief. I will upgrade the rest later …

You are wondering by now when I am getting to the test … a few moments more. One thing that has grown tiresome over time is the context switching between the laptop (which I keep open on one monitor through remote desktop), my main machine for research (which I do a lot of when I am working as I choose to look up issues and questions rather than trying to write down every detail) and the virtual machine.

So I need a third (and eventually a fourth) monitor. This would allow me to arrange each of the contexts on a separate monitor and thus I would be able to work without constantly flipping windows. It is an efficient way of working and is essentially irritation free.

Except that my GTX570 will not support more than 2 monitors, despite having many extra connectors on the back. But the newer GTX650ti or GTX660 will do that just fine. And they have twice as many cores, so my Sony Movie Studio rendering times should shrink, yes?

No. In fact, the CUDA architecture starting at the 600 series cards is called Kepler. The architecture in the 400 and 500 series was Fermi. And guess what … Fermi is much faster for compute operations, although Kepler is much faster for pure gaming. So just swapping out a 600 series for my 500 series is not obvious … yet there are reports that some people have got rendering times that are similar. And I thought … maybe the 660 would work as well, or maybe GPU rendering is not such a big deal in Movie Studio anyway … after all, it’s not Vegas Pro.

Well, in fact the GTX570 speeds things up quite a bit. I tested it be re-rending Nick’s cover of “I Killed the Monster” in each mode and the times were pretty dramatically different. I used identical settings in each case and even used fairly light weight settings like 720p. Still a big difference …

Now, part of this is the fact that the file is much larger for the CPU because it rendered a slightly softer interpretation of the shadows, bringing in a lot of extra shadow detail that it had to process. And the file it produced is significantly larger, despite the actual detail being essentially identical to the eye.

CPU vs GPU

Also interestingly, I prefer the slightly more contrasty look that the GPU rendering chose and I prefer it. So 30% shorter rendering times and 18% smaller files with a nicer contrast (at least in this case) and no discernable difference in motion or detail (again, for this example) makes me think that GPU rendering matters a lot to me.

Which means that jumping to a Kepler card should probably wait for a while.

So how so I get that extra monitor if I don’t want to grab a card that supports the extra card? Well, I have a few choices. One is to try adding one of my older video cards into the box. For example, I have an older GTX9600 … which even has some CUDA cores to add to the party. So I may give that a shot. (Note: I already tried it once but screwed something else up and it would not boot, but it turned out to not be the card’s fault as it continued after I removed it. I’ve fixed that so I plan to try again.)

If worst comes to worst then I can always just buy a cheap USB video card to get that third monitor for now. There are many options … but for those who use Movie Studio or Vegas Pro, the GPU rendering definitely beats CPU rendering, at least for my CPU.

My CPU is the 8 core AMD8150, which is running at the stock 3.6ghz (turbo 4.2ghz.) It is a bit slower to render in the Adobe Premiere video benchmarks on Tom’s Hardware than the 2700K Sandy Bridge, but is in the ball park. The extra cores help of course. The ratio of time to the fastest processor on the planet at this time (i7-3930K 6 cores / 12 threads) is 252/173 or about a 29% difference. Since my GTX570 already gives me more of a difference, I think I’ll just stick with that.

Anyway … for anyone out there who desires fast rendering, the GTX570 and 580 cards are a good bargain right now. Were I do jump to a Kepler card, I think the lowest I would consider is the GTC660ti. That  has enough cores to overcome the lack of agility in the Kepler architecture for rendering. YMMV of course …

If you are still wondering why we need a change of government at the federal level …

Yes, that’s right … “vigilance” over our health products in Canada is overseen by a committee that has a 3/4 majority of members with a financial interest in the outcome.

One wonders then how much chance there is of policies that promote health over financial interests (duh :-) …

The “cronies and corporations first” and the “deregulate or pander” policies of typical conservative parties in North America has crushed the poor and middle classes while opening the door to disasters across the health related industries of food and environment.

How can this be stopped?

By voting them out and hoping that the other party cares even a smidgen more …

Thanks go to Joni Freedhoff at Weighty matters for this article.

Battle of the Newest micro four thirds (m4/3, MFT) Sensors at 3200 ISO – Who is the high ISO champion between the Panasonic G6, Panasonic GX7, Panasonic GM1 and the Olympus EM1?

There are four very recent m4/3 cameras that are of great interest to buyers. They are not comparable when it comes to size, weight, specific features, etc … but they are definitely comparable when it comes to image quality. Note that I have little patience to people who speak about JPEG engine comparisons because that’s like comparing million dollar sports cars by how smooth the ride is on the way to the grocery store on the corner. Advanced users just don’t worry about such trivialities as they want to know what they can wring out of the cars …

So I always compare what can be done in raw, because that is where the advanced user is working. It takes no more time than processing a JPEG and the results are often far superior, depending on the user’s skills … but the point of doing advanced work is to develop the skills to wring the best you can out of the camera, in this case. Otherwise, just take it back and forth to the grocery store and be done with it. So … if you are interested in how the newest m4/3 cameras compare at 3200 ISO in raw then read on.

The four newest micro four thirds (as mention already in the title) are the three new cameras from Panasonic – G6, GX7 and GM1 and the newest from Olympus – E-M1. The E-M1 has a new sensor that is similar to that of the E-M5 but with pixels for a form of phase detect auto focus (PDAF.) I am not covering anything about the bodies and features, as the variance is significant. You can find myriad reviews out there, some of which don’t necessarily completely gloss over what is important :-) so have fun …

Now … I don’t own any of these, and I usually compare crops from imaging resource, but they for some reason are not shooting the mannequin series for all the cameras these days. So I will do that one later. DPReview have a new set of comparisons and they don’t want anyone to reproduce their files, which I will of course respect … but I will show small crops from their files after I process them so as to explain what I am about to say about a direct comparison.

There is no significant difference in the high ISO performance in raw between any of the 16mp sensors. I’ve published many an example of this and I continue to believe it, since I can find no reason why I would shoot one of them over another. JPEG engines are a different story, but even then I do not favour the over-processed Olympus JPEG look, but I recognize that they have had a better JPEG engine until the most recent Panasonic generation.

Anyway, back to raw comparisons … the only exception I make to the 16mp sensor rule is the GH2, which has some issues in shadows. Panasonic created that as the very first generation of sensor and it was and still is a magnificent video sensor and a decent stills sensor. The version of that sensor that found its way into the G5 and G6 is much better for stills and still amazing for video. But, of course, if you are on a budget for a video and stills camera, then the GH2 remains an excellent choice since it has the ability to run very high bit rates after a simple hack is installed. I run Cake 2,3 and get 64Mbps in low light and 60 outdoors … where the subject is smooth, I still get 46Mbps … this is far above what you can get in AVCHD stock on any camera, including the GH3.

So anyway … back to the four amigos … I find that comparing raw images on the DPReview web site using their excellent comparator is less satisfying than it used to be because a lot of the details are kind of small. And of course the raw settings they use are pure defaults as they must be. So they answer the question “what do images look like if I never touch the settings on my camera?”

That questions interests me about as much as “what would my house look like if I leave the contractor’s white wash on all the walls?

So the question I always try to answer is “how similar can I make these look at the best settings I can find?”

That question allows me to decide if there is one sensor that is superior to the others. And guess what … I don’t think there is. The colors look quite similar to me and are of course infinitely variable in Lightroom anyway, so you could do whatever you like with them. The tones all look very similar in this generation of sensors, despite the fact that some of these are famous for being amazing and others are descendants of the earliest 16mp sensors. And noise seems about equal across the board. But then, I never miss an opportunity to make that point, do I :-)

So … the crops … I will leave it for you to look at the comparator to see what it looks like on DPReview’s site. Here is a small screen shot for you … and it links directly to their site. So go exploring there at some point …

So the processing of these images was to use the white balance as set on one images to make them all look about the same, and to use the same tone curve on them all. That they respond so similarly is a credit to the latest generation of sensors. They all respond alike. I used identical amounts of noise reduction (a small amount of luminance NR to smooth the grain and the default amount of color NR) and the set sharpening at a small radius with the edges only (by using some masking in LR.)

The one area where I tweaked things was to make the images look about equally sharp and detailed. This required a small tweak in some instances, and the difference is essentially attributable to slight differences in the strength of the AA filters used in these cameras. Remember that every one of them uses a different sensor assembly, and it is unlikely that they all use identical AA filters, although that is possible between the GX7 and the GM1, despite having what appear to be different underlying sensors.

Note further that I am prone to leaving a small amount of grain in order to retain the highest possible level of details for these sensors at this high ISO. If displaying on the web, the downsizing will wipe out the grain and if displaying on paper, the ink bleed will do the same job. So this amount of grain is almost never relevant.

Crop 1

A slight difference in the darkness of the background. On my IPS monitor, the background is smooth and quite dark, on my TN panel (which has a softer tone curve) the background is slightly grainy and mid toned. The color and detail in the actual subject figures is essentially exactly the same. If you see any of them dominant then I suggest that you stop reading here as your biases have already won :-)

Crop 2

All four are grainy. The EM1 might be slightly less grainy but the difference is so subtle that it means nothing much. In fact, on the IPS panel, the difference is entirely invisible (and since that is the panel that can be and is properly hardware calibrated, that is how I come by my opinions.) If you see a massive difference and you have an uncalibrated TN panel, then I suggest that you consider the effect of bringing a water pistol to a gun fight. Because that’s what you are doing when you try to form opinions on image quality on your monitor …

Crop 3

The color response of the G6 is fractionally different I think in the darkest colors here. But It is far too subtle for my eyes. The detail it delivers is, however, the same or even slightly better than the others. These sensors are all magnificent.

Crop 4

The first real weakness we see is in the edge detail of hair at this high ISO. But they are so close to one another that I fault them all equally … and not very much at all. For example, if I had included a NEX APS-C sensor in the mix it would almost certainly be worse than all of these as I have shown in the past. Sony’s sensors simply do not compete, even in raw. In my opinion, at least. Note how essentially identical the line patterns look. There is no relevant difference in how these sensors respond to sharp lines.

Crop 5

Finally, a human head. :-) I’d say that the micro contrast is so close that I fail to see why anyone would not be happy with any of these. I keep wanting to think that the G6 is somehow slightly inferior, but once you examine the top row and the bottom row at exactly the same spot on your monitor (I’m serious) any difference in micro contrast vanish. At any reasonable display size, material differences do not exist.

Crop 6

I spent a lot of time trying to understand the differences here. The bottom of the two smaller middle circles shows a light highlight in the right middle on the GX7 crop that is not present on any of the others. The top right circle has darker patterns on the bottom pair than the top pair. The interference patters are similar, but have subtle differences. In the end, they are all surprisingly detailed for this ISO and none looks better to me than the others. Obviously, there will be some interactions between colors at the edges that will differ slightly depending on the specific sensor you buy. But I don’t think you can say that one will be better than the other.

Conclusion

So what can we possibly say about these crops? My opinion remains that you should not spend you time trying to look at image quality differences. It all comes down to what matters to you in performance characteristics.

So which one should you buy? That’s actually not very difficult …

• If you want pocketability then the GM1 is a no brainer from this group. I regularly carry the GF3 and 14-42 X pancake in a jacket pocket, and this new camera is smaller with a much better sensor. Alternatives include the GX1 and GF6, which share a terrific 16mp sensor with similar image quality and are a lot cheaper.
• The three Panasonics all have excellent video. It’s a thing with Panasonic, so if video is your think then you should just look at the various Panasonic bodies. Alternatives include the GH3 if you want the very best video available in this class of camera. The GX1, GF3 and G3 can all be hacked, as can some of the earlier bodies. They deliver very nice video on the cheap.
• If you want IBIS then you are looking to choose between the EM1 and the GX7, both of which are expensive, although the EM1 is really, really expensive lol. The IBIS on the EM1 can be used during video, do if you want to shoot casual video with older fast primes, then the EM1 is a no brainer. On the other hand, the GX7 has focus peaking (as to all the Pansonics here) so if you don’t mind stabilizing the camera externally, then the Panasonics will be easier to shoot in focus.
• If you want a professional feeling build, then the EM1 is a no brainer.
• If you are on a budget then the G6 is almost a no brainer with the GM1 right behind it. In fact, with a lens, they come out about equal in price and much cheaper than the other pair.
• If you like the dSLR feel then the G6 is a no brainer.

And so on and so forth. One thing is certain … if you cannot get good images from any one of these cameras, then you need a skill upgrade. And that goes for any m4/3 camera. There is not a bad one in the bunch for shooting in good light, although I would probably not bother shooting any of the 12mp sensors at 3200 ISO or above.

It can be done, but you might need to process for black and white as you can see in this example of the GF3 in very bad mixed low light at 3200 ISO …

How much does your choice of lens affect your videos?

This is a little test I performed in answer to a thread on http://dpreview.com in the micro-four-thirds group. In that thread, someone had used a prime and a kit zoom on a Panasonic GH3 (the premiere mirrorless camera for video) and found that the noise dramatically increased with the kit zoom, despite having not changed any of the settings.

Of course, in the end this is not really possible since the same aperture would return basically the same results. However, there will always be some difference in the micro-contrast and sharpness of each lens, which will definitely make an impression in the final results.

I ran the test in brutally bad light … a pair of really old halogen lights (more than 2 decades old) pointed at the ceiling more than ten feet away on camera left and 8 LED bulbs coming through a bathroom door about 8 feet away at camera right. The video exposure at 1/25s and f/5.6 required 5000 ISO, which is pretty high. Still, you will see that the Panasonic GH2 does a beautiful job at that ISO, possibly because I am running the Cake 2.3 patch, which allows the camera to go – in this case – up to 64Mbps, which is almost as high as the best possible setting on the GH3 and far higher than any other consumer camera will do natively.

I used f/5.6 so that all the lenses could shoot the same aperture, but do note that 5.6 is wide open for one of the kit zooms, and I think it shows in a slight loss of contrast. I graded the entire video track in Sony Movie Studio Platinum Suite 12, raising the brightness and contrast to emphasize the noise. I find the result excellent … strong saturation and very tolerable noise that would vanish if I owned Neat Video or its equivalent.

So … the test. Tripod mount, GH2 set to 1080/24p and the highest quality setting with the Cake 2.3 patch. At 5000 ISO the average bit rate is 64Mbps and at 400 ISO the rate drops to 46Mbps because of the lack of noise and details (much more background blur) to encode. The four lenses in order are: Panasonic 14-140 G Vario Mega-OIS 4.0-5.8 (the mark 1 version), the Panasonic 14-42 X Vario Power OIS 3.5-5.6, the Olympus 45 1.8 and the Pentax M (I say K in the movie but I think it is the M) 50 1.4, which is of course manually focused. I set AF continuous, and none of the lenses pumped focus, and none of them showed any jitter despite me leaving OIS engaged.

My opinion of the results is exactly as I describe above. You see no difference in noise, but you do see a difference in magnification (unavoidable since the lenses are not quite identical in focal length) and in contrast and sharpness. The primes look better, but only because you see them all one after the other. Otherwise you would not notice anything amiss in my opinion.

What I think is really illustrative, though, is the drop from f/5.6 to f/1.4 … an astounding 4 stops. The ISO drops also from 5000 to 400 … and the results are just so pretty :-)

I hope you enjoy it …

Gamla Stan, Stockholm

Gamla Stan stands for “Old Town” as it remains the original settlement that became Stockholm. It dates back to the 13th century and is a really enjoyable place to walk at any time of the day or night. I’ve visited Stockholm four times on business and have managed a lot of hours in the evenings and a few Saturdays (stay over one Saturday and you cut your air fare in half on Air Canada.)

I am going to show a single image from an evening a few weeks ago when it was raining. We took the subway from the suburb Kista and transgered at T-Centralen (Central Station Tunnelbana) to head southward one stop to Gamla Stan. We walked to a restaurant (C+C I think it is called) we had been wanting to try since I had seen it 6 weeks earlier on my 3rd visit to Stockholm. We eventually found it and I had an amazing plate of Swedish Meatballs with potatoes and lingonberry sauce. Anyway, it was still raining when we got out so everyone scattered to their hotels (four of us from Linköping, Malmö, Boston and Ottawa stayed in different hotels, but that’s not much of a hassle in Stockholm as the subway is excellent.) I stayed on Gamla Stan though and shot images for another hour or thereabouts …

So this is the image I like the best from that evening …

I liked it enough that I entered it at http://panachallenge.com, where it took second place in a challenge called “Wet” … garnering 6 first place votes, twice what the 1st place image managed. However, the winner gathered more seconds and thirds, and even though we had the same number of top ten votes, the weighting system was enough to boot my image from first place. Someone in the comments suggested that I don’t wait so long again before entering (it has been years), but this kind of result is exactly why I don’t enter contests very often. The winner is a nice enough image, but it’s just another flower with a few droplets on it … sigh.

Anyway … back to Stockholm. Sweden is the kind of place that a Canadian could find very homey … everything there feels somewhat familiar to me. The rugged landscape reminds one of the Canadian shield, which covers a good chunk of Ontario and nips into Manitoba too. Hilly with lots of lakes and trees, the two countries share an outdoor culture that really seems similar to me. They hunt and eat moose too :-)

I like this image enough that I did a somewhat high key black and white version to frame … it looks like this:

Not sure how it will turn out on matte paper, but if it looks nice enough I think I will try it as a canvas.

I’ll try to process my multi-thousand pending images over the next months and post a few more observations of Stockholm and Linköping … I certainly recommend Stockholm as a nice holiday destination. I would love to visit one day around the midsummer festival. It’s supposed to be quite something …

Snow Tire Time … the last warm sunny day comes out of nowhere to make it really easy :-)

But there was a cloud on the horizon … the rinky dink toy hydraulic jack that I have used for years has been a pain in the butt for equally as long. It hates lifting the CRV because it needs to sit on 4 inches of platform just to be able to reach. And today this contraption tipped over and the CRV rolled off of it.

That was it as far as I was concerned, so I dashed off to Canadian Tire and bought the real deal. A 2.25 ton (3 is better, but none in stock) monster designed for low profile cars (which really came in handy when I tried to sneak it under the Mazda’s skirts) … it dwarfs the lowly toy as shown here:

f770exr  iso125  f/3.5  1/30  4.6mm (24mm efl)

I changed the CRV’s tired in record time, since I was only moving the jack around and not the dang platform (which consisted of a pair of chunks of 2x10) …

And the Mazda went even faster …

And in a fit of clear thought (so rare these days :-) I switched from the air gun to a tire gauge / filler attachment and set all the tires on both cars to the recommended values (26psi for the CR-V and 35psi for the Mazda 3.)

I am always pleased when I don’t actually waste the whole weekend :-)

By the way … have you got your snow tires on?  It’s time …

Published in the Ottawa Citizen again …

My thanks to Robbi Hay, editor of the Ourtown section of the Ottawa Citizen for using one of my images this week as the section header … I shot this years ago along Fallowfield Road with the Nikon D2Hs, an excellent 4Mp professional APS-C dSLR …

Head out this Thursday to grab your copy and see what is on this week in Ottawa …

Rogers cable … still the king …

I am a little frustrated by the huge leap from the plan I am on – extreme plus I think it is called – to the one I want – ultimate. The cost is almost double, but the speeds are almost triple, so that sounds like a bargain. Still, as I have mentioned in the past Rogers has the best technology for now, and this peak time speed test kind of confirms it, at least anecdotally.

PCMag tested all of the ISPs in Canada and Rogers came out as #1, which did not surprise me.

However, there is a small cloud in that victory, which is to say that people like me who are on the cusp of transitioning to the highest speeds have a somewhat unreliable modem from Rogers – the SMCD3GN, which is the standard DOCSIS 3 modem and router that they have used for years.

The modem most favored by pirates (get IT?)

This modem is not quite reliable as a cable modem and is abysmal as a router. Make sure that you acquire a good router (I love the Apple routers) and that you ask Rogers to put your modem into bridge mode to disable its internal routing function. This makes a huge difference.

But I am about to embark on a quest to get them to give me a CISCO modem instead, as it is universally lauded as the most reliable as a cable modem. Its router is also a little weak and I certainly would not use it after experiencing the nirvana of the Apple Airport Extreme … but YMMV.

Rogers Cable – bug in the web site ...

The following bill makes it look like their computer cannot add ...



That’s right … 60 bucks (mine is an “all fees included” price, or so I was told) minus my bundle discount should equal $52.80. But apparently it equals $61.59, upon which taxes are applied.

I called and it turns out that the PDF is the official bill, and it shows the rather high price of just under 9 bucks for the call display bundle ... sheesh. This summary page is neither official nor accurate. Hopefully they will fix this soon as it is really misleading and makes their web designers look really foolish.

Cable providers like Rogers continue to make really cynical moves …

This one really chaps my ass …

I have no desire whatsoever to subscribe to get movies. I get more than enough on Netflix, for which I happily pay every month.

But … Rogers knows how valuable HBO is as one of -- or arguably the -- premium cable channel(s) with original programming, so they make this expensive movie package the only way by which a subscriber can get this one completely unrelated channel.

Yes … our government (Canadian Radio and Television Commission) is once again asleep at the switch. This is a cynical cash grab by Rogers and there is nothing the consumer can do except pay through the nose to get access to this one channel. It is not available as a single channel.

The other incredibly cynical thing they do is to not make your info available in a consumable format. There is no summary of your current channel lineup and no way to just review your options. Instead, you have to click on buttons like “add more channels” and anyone who knows Rogers quakes in fear at the thought of accidentally authorizing them to rob you blind. What pops up is a live chat request.

You read that right … there is no transparent view of what you are paying for. You have to ask a Rogers agent for information about what you subscribe to.

Surely this lack of transparency should be illegal?

I remain convinced that they have the best technology currently available and as consumer unfriendly a set of practices as exists in the industry.

Of anal probes and stop and frisk …

New Mexico … a man leaves a WalMart and rolls a stop sign. The police stop him and he gets out of the car, apparently clinching his butt cheeks. This is kind of understandable as a nervous reaction.

So the police suspect that he is hiding drugs up his butt and they get a warrant to have him examined. The first hospital declines on ethical grounds so they find another hospital, where he undergoes xrays and probes (more than one) and eventually gets sedated and colonoscopied (yeah, I know that’s not a word) …

So New Mexico is taking a page out of the easy coast’s shameful stop and frisk policies, but are renaming it to stop and fist ;-) ... Obviously, they are having some trouble with the concept.

Seriously … things like this seem to be happening with disturbing frequency in the USA. Watch the video … they mention other recent violations of personal rights and freedoms. They even use the term “police state” to describe the mood in the US recently, and I find that terribly sad. How could a beacon of freedom fall so far so quickly?

The Canadian Heart and Stroke Foundation Health Check Program has lost its way …

They have been putting their stamp of “goodness” on more and more sugar laced foods. It seems that their motivation is purely monetary, since there are products that they endorse whose calorie count is 96% sugar, such as the subject of the video below by Joni Freedhoff, head of the non-surgical bariatric unit at the Ottawa Hospital.

Registered Dieticians are supposed to be sensitive to sugar, as there is a diabetes type II emergency in Canada and in fact world wide. Yet they continue to endorse this.

The penalty we will all pay for the insidious nature of the relationships with big food is the continued declining health of our population as these children – who will grow up thinking that candy is fruit – become diabetic even earlier than previous generations, who at least knew how harmful sugar was …

Movi M10 Stabilizer … 15 grand of goodness …

I was poking around today, looking at some tools I would like to acquire at some point (ACID Pro 7 and Vegas Pro 12 Edit for example) and one thing led to another and I found an unboxing video for this stabilizer. Stabilizers have always interested me because I like farting around exploring hack amateur level video. As a dilettante enthusiast, I find the one truly difficult thing about video is getting stable shots when you are moving.

But I am not alone. Professionals suffer from this issue too … at least at the indie level, where budgets are pretty tight and the crew cannot number in the thousands. The linked video below in fact is made by an indie team, where the short they were making is shot 100% with this stabilizer.

The M10 is a pretty recent arrival (I think it started shipping this year in mid-August) and wow … it solves a whole lot of problems. It has gimbals that run off of battery, so it plays an active role in stabilization. And it has remote control, which allows two operators to split the tasks of camera position and framing. Of course, it is able to work well with one operator too … so the options are almost limitless.

Anyway … I found a fantastic video showing how it works with clips behind the scenes and the actual resulting clips showing how stunningly well it stabilizes motion … I find it breathtaking. And the cost is actually pretty reasonable when you consider what it would cost to attempt these shots with gear instead of humans.