AF during video, comparison GH2 vs GH3

Recent Micro Four Thirds cameras have very good autofocus performance for still images. Mostly, the performance is among the best in this class, certainly better than DSLR cameras in live view mode. However, there is one area where mirrorless cameras don't perform well at the moment, and that is continuous autofocus: Both during video recording, and for photographing moving objects, e.g., for photographing sports and birds.

Some camera manufacturers have been trying to solve this by adding phase difference sensors (PDAF) on the imaging chip, like the Nikon 1 and Canon EOS M cameras. However, the real world benefit of that solution is still undecided.

Panasonic have said in interviews that the on-sensor PDAF solution is not going to be used for Micro Four Thirds, at least not anytime soon. Rather, Panasonic expects to achieve better continuous autofocus performance by using faster image readout from the chip, better image processing algorithms, and more processing power. Have they achieved this with the most recent Panasonic GH3?

To compare the autofocus performance during video recording with the GH2, I used a Lego Technic contraption to move a cardboard box back and forth at a steady pace. I then set up both cameras, in turn, with the Lumix X 12-35mm f/2.8 lens at 35mm f/2.8 at close range, and recorded video at 1920x1080, progressive, 25fps. Comparing the resulting footage, it is easy to see which camera better finds the focus during the movement. Here is the video footage, for comparison:



It's easy to see that the GH3 achieves correct focus more often than the GH2, in fact, about twice as often, according to my frame counting. The GH3 also has a better overall sharpness: It is possible to see the offset printing pattern more easily with the GH3. This could be partially due to the multi aspect sensor feature that the GH3 misses: It means that you get slightly narrower field of view when using the GH3, as compared with the GH2, and hence, more enlargement of the subject.

One of the new features of the GH3 is the 240fps contrast detection autofocus (CDAF) sensor readout. The GH2 only does 120fps, maximum. The smaller print in the GH3 specifications state that the 240fps is only possible when using the newest f/2.8 zoom lenses, the Lumix X 12-35mm f/2.8 and the Lumix X 35mm-100 f/2.8.

And can the camera use the 240fps feature during video recording? Probably not, since the sensor is busy reading the image at 25fps for the video stream anyway. I previously compared the GH2 video AF performance during 25fps and 50fps video, and found that it does better at 50fps, indicating that more frequent image readout is better for the AF performance. As long as the shutter speed is faster than the video rate, there is surplus time between the frames for CDAF readout. Perhaps the GH3 camera can utilize this for better AF performance?

Conclusion

The Panasonic GH3 camera appears to be able to focus better during video recording than the GH2, even at the same frames per second (fps) rate.

Timelapse with the GH3

One of the new, fun features of the Panasonic GH3 is the timelapse mode. Sure, one could do the same thing with previous cameras as well, given that you buy an external controller to connect to the camera's remote shutter release socket. But having the timelapse possibility built into the camera is easier.

To make an example timelapse video, I first set the camera on a tripod over my table, like this:



I'm using the Manfrotto 190XPROB tripod, which is useful since the column can be set horizontally. The ball head is Benro B-2, but most ball heads can be used here.

I used the Lumix G 14-42mm f/3.5-5.6 kit zoom lens at 14mm, and set the aperture to f/3.5. This was done mostly to avoid the aperture closing and opening for each shot, which is a somewhat annoying ticking sound. I prefocused, and then set the camera to manual focus (MF), to avoid the camera engaging the autofocus mechanism for each shot. I also used the electronic shutter mode of the GH3.

GH3 electronic shutter

I've written about the electronic shutter feature of the Panasonic GH3 before, concluding that the feature is very useful, but has some side effects. The electronic shutter is useful since it is totally silent, and vibration free. However, the electronic shutter reads the image sequentially, row for row, rather slowly, taking about 1/10 second in total. If you move the camera during this time, you get odd wobbly effects, even with a very fast shutter speed. This is a rolling shutter artefact.

To illustrate this, I have superimposed two exposures in an animated GIF. They were taken with the Lumix G 14mm f/2.5 at f/4, 1/60s, ISO 200. Normally, one would think that 1/60 second is safe for handholding a shot with a wide angle lens, however, since each full exposure takes 1/10 second, any movement during the exposure will result in a skewed image:



Looking at only one of the two exposures above, one might not notice any problems. However, when seeing both, it is clear that at least one of them, probably both, are not geometrically correct. So, is this a problem? When holding the camera reasonably still during the exposure, and not photographing very square objects, it is no issue. If you critically need rectilinear images, then you are better off using the normal mechanical shutter.

ETC comparison, GH2 vs GH3

I recently looked at the video quality when using the ETC function, as opposed to using non-cropped video on the Panasonic GH3. My conclusion was, perhaps not surprisingly, that the ETC function adds quite a bit of noise at higher ISO. So while the ETC function is very good to have for extending the reach of your lenses during video mode, you should avoid using it at high ISO unless you really have to.

But what about comparing the GH3 with the predecessor GH2. Both cameras feature the ETC mode. To test this, I set up both cameras to video record the same scene, and used the Lumix G 100-300mm f/4-5.6 lens at 100mm f/5.6. I already know that this lens is very sharp at 100mm, especially when stopped down to f/5.6.

Here is a comparison of the videos outputs:



Also, here are some 100% crops from the video streams:



I think that the cameras perform quite similarly, in this case. I'm actually thinking that the GH2 videos look better in terms of sharpness and noise at ISO 3200. The GH3 has better colours, though.

Of course, the GH3 goes all the way up to ISO 6400 for video, also for ETC, which is an advantage over the GH2 anyway.

One can also argue whether ISO 200 is the same with both cameras. Do the cameras have the same sensitivity at ISO 200, and at higher ISOs? I don't think the jury is fully out on this yet. To try to assess this, I tried to take images at ISO 200 with the cameras GH1, GH2 and GH3. However, even after seeing the results, I am not sure how to interpret them.

Conclusion

The ETC mode is very useful, but avoid using it at high ISO, where it adds quite a bit of noise. The GH2 and GH3 appear to be quite similar in this respect.

Using the Lumix G 100-300mm f/4-5.6 at 300mm in combination with the ETC mode gives an effective reach of around 1500mm equivalent, filling the moon into almost the entire video frame:

Tele lenses and perspective

One can often hear phrases like "tele lenses can be used to compress the perspective, and make things that are far apart appear close". But it is true?

As the statement it written, it is wrong. A simple geometric argument shows that the perspective only depends on the distance from the camera to the subject, regardless of the field of view, regardless of the lens.



But why believe a geometric argument? We can look for ourselves at a real life example. This image was taken with the Lumix G 100-300mm lens at 300mm:



In the image, it looks like the construction cranes are just behind the neon "Freia" logo on the front rooftop. But in reality, the distance is 1 km. This is what we normally call a compressed perspective.

But is it the long focal length which produces the compressed perspective, as the statement initiating this article? We can see for ourselves, by setting the lens to 100mm, and photographing from the same point. That gives this image:



Does this image have a different perspective? We can crop the centre of the 100mm image and compare it with the 300mm image:



We see that the images are exactly identical. There is no difference in the perspective, demonstrating that the perspective only depends on the placement of the camera, not the focal length.

On the other hand, if I change the point from which I am photographing, as well as change the focal length, then I get a totally different perspective, as illustrated:



Here is the same building photographed much closer with the Lumix G 14mm f/2.5, a wide angle lens:



Now, the perspective is not compressed anymore. But was this due to the wide angle lens? No, I would say, it is because I went closer to the subject.

But, it was the wide angle lens which allowed me to go closer to the subject. Just as it was the tele lens which allowed me to go far away from the subject, to create a compressed perspective.

So while it is not true to say that a tele lens compresses the perspective, it is true that the tele lens allows you to step further from the subject, to achieve a compressed perspective. So, at the end of the day, it is perhaps not so wrong to say that the tele lens compresses the perspective, as this is what you achieve when using the lens anyway.

Conclusion

Having some focal length range available, either as a superzoom lens, or as some different prime lenses, is useful. It allows you to achieve various perspectives, by varying the distance to the subject.

GH3, quality of ETC video vs non-ETC

Just like the predecessor GH2, the Panasonic GH3 has the ETC (Extended Tele Conversion) mode. This is useful for videos, when you need more tele effect, a longer reach.

Normally, the camera uses the whole imaging sensor during video, and scales the output down to 1920x1080 pixels for the video stream. In ETC mode, though, it only uses the central 1920x1080 pixels of the sensor, giving an effective 2.4× crop factor, while retaining the full resolution, see the image below:



With the Panasonic GH2, the ETC crop factor was 2.6×. The reason for the difference is that the GH2 has an oversized, multi aspect sensor. The GH3 does not.

Let's say that you use the Lumix G 20mm f/1.7 pancake lens. When using ETC during video recording, it effectively becomes 48mm f/1.7, with 48mm = 20mm × 2.4. Which can be useful at times.

The ETC mode can be enabled in the video menu (left below), or by using the Q-menu (right below):


This feature is very good to have when you want to record videos at a long tele, and you don't have a long enough lens. However, since the camera has fewer pixels to use for making the video stream during ETC mode, one can guess that the quality will suffer. Unlike when using the full sensor, there is no possibility to scale down the image for better noise performance.

Comparison: Non-ETC video vs ETC video

To compare the non-ETC output with the ETC output of the GH3 camera, I recorded video sequences using the Lumix X 12-35mm f/2.8 lens at 35mm without ETC, and at 14mm with ETC. Both these two modes give approximately the same field of view, since 14mm multiplied with the ETC crop factor 2.4 gives approximately 35mm.

I used f/5.6 for the best sharpness. In both these modes, I used ISO values from 200 to 6400. I used the 25fps 1080p, ALL-INTRA mode for the best video quality.

Normally, one would of course not use the ETC mode with the Lumix X 12-35mm f/2.8 lens at 14mm. It would be better to just zoom the lens to 35mm. However, using this trick, I was able to compare the non-ETC and ETC modes using the same lens.

Here is a video summary of the comparison:



And here are 100% crops from the videos, for comparison. The normal non-ETC images are on the top, and the ones with ETC on the bottom:



We see that even at ISO6400, the non-ETC video footage holds up pretty well. It is a bit noisy, and the colours are not as good, but the video is still usable. With ETC, though, the quality degradation is much more severe.

Conclusion

The ETC mode is a very useful feature, but it is best used at low ISO. At higher ISO, the video quality will suffer. We also see that even at ISO 6400, the video quality is quite good with the normal, non-ETC mode.

We don't know exactly what algorithm the GH3 uses for scaling down the video stream from the full sensor to the 1920x1080 pixel output. It probably does not use every 11 million pixels of the sensor (in the 16:9 subsection of the sensor area), that would take too much processing power. But it is clear that it does use some averaging technique, to keep the noise down. In ETC mode, there is no scaling down, and hence, no noise reduction from using more pixels.

I've also compared the ETC video quality of the GH3 with the predecessor GH2. They seem to perform quite similarly, although I generally like the colours of the GH3 better.

When using ETC with a long lens, you can get an extreme tele reach. In the example below, I am using the Lumix G 100-300mm f/4-5.6 at 300mm, together with the ETC mode. This gives a very compressed perspective, due to the extreme tele effect.



You'll see the image wobbling. This is due to atmospheric disturbances, since the light travels through a lot of air before it reaches the camera. There is no way to avoid this, beyond, perhaps, getting up early in the morning while the air is cooler.

When using the Lumix G 100-300mm lens at 300mm, and with ETC, the effective equivalent focal length becomes 1500mm, which is a lot. Even when placing the camera on a tripod, I had to remove the first seconds of video footage while waiting for the camera and lens to settle down after pressing the shutter button.

New firmware for GH3

Panasonic has announced that a firmware upgrade for the Panasonic GH3 will be available by the end of March. The firmware upgrade improves the autofocus performance on some newer zoom lenses, like the Lumix X PZ 14-42mm and Lumix X PZ 45-175mm lenses. It also makes it easier to connect to Wifi when using Mac computers, and adds a video mode.

When the new firmware is released, some people will try to decipher the firmware, and try to make changes to it. Most likely they will not be able to add genuinely new features, but perhaps they can make changes to parameters like the video bitrates. This is usually called "hacking the firmware". The GH1 was popularly hacked, giving the possibility to use a higher video bitrate.

Why use RAW?

You'll often hear people recommend to use the RAW image file format. But why? What is the reason for using the RAW file format? In this artice, I'll look at an example where it makes a difference.

Most serious cameras have the possibility to record the images in two different file formats: JPEG and RAW. The RAW file format can be of many different types and extensions, but they all have some things in common: The RAW files take up much more space than the JPEG file, and they contain the image as seen by the camera sensor, mostly without any adjustments.

This is in contrast with the JPEG files, which are compressed to take up less space, and where the camera has tried to interpret the image how it thinks you want to see it. With the RAW file format, you can later develop alternative JPEG output files by altering how you want the original RAW file to be interpreted.

Here is an example JPEG image. It was taken with the Panasonic GH2 camera, with the Lumix G 14mm f/2.5 lens at ISO 640, f/2.8, 1/10s.



The shutter speed is too slow for taking a picture with this lens, according to conventional wisdom: To be reasonably sure to avoid camera shake blur, one should use around 1/30s or faster with a wide angle lens like this. However, in this case, I preferred to use a slow shutter speed, and avoid using the flash or a higher ISO. Using the flash would have changed the style of the photo totally. Besides, I like the effect of the slow shutter: The trailing lights from the sparks coming out of the pipe can be seen. And the image is still reasonably sharp.

The picture was taken late dusk, which is what gives it the blueish tint. The blue colour is correct: The scene was in fact blue due to the timing, just after sunset. However, by using the setting "Twilight" in the RAW developer program Silkypix, I can get more natural looking colour tones:



This image has a totally different appearance, with more normal skin tones, and less blue.

One could also go the other way, and exaggerate the blue dusk colours, by changing the white balance and increase the saturation:



Some would say that the colours now look unnatural, others might prefer it for the stronger colours. But the point is that when you have the RAW image, you are free to make these choices later. With only the JPEG image, it is more limited what types of adjustment you can make, without significantly losing colour details.

With that said, you are less likely to want to make colour tone adjustments to images taken in daylight, with a proper exposure. In those cases, you are often well served with the JPEG image only. In my opinion, the RAW image is of more use when the there is dim light, and when the exposure is uncertain. In sufficient daylight, with normal contrast, the camera often does the right choices with the out of camera JPEG image.

Pentax K-01 is history

Pentax has now ceased the production of their only large sensor mirrorless camera, the K-01:



The camera was in production for about one year. It was without doubt the most unusual mirrorless camera of all, for a number of reasons.

First of all, it is large. The main reason is that it does not employ a shorter register distance. When removing the mirror assembly, the other mirrorless systems also shorten the register distance, allowing for a smaller camera body. Also, the shorter register distance allows for constructing smaller lenses. Especially wide angle lenses can be made smaller this way.

Pentax's choice of retaining the same register distance and lens mount as their DSLR cameras has a clear advantage: They don't need to create a whole line of new lenses. Hence, their investment is much smaller than that of the Micro Four Thirds and other mirrorless systems: Panasonic and Olympus needed a whole array of lenses to be a viable alternative to existing DSLR systems.

While the camera was launched together with a new 40mm pancake lens, this lens is based on an existing lens design for their DSLR line of cameras, and, hence, is not really new.

Also, the Pentax K-01 is a very unusual looking camera. Designed by Marc Newson, it is a camera you either lover or hate. The camera probably lets form go over function, but still is quite good to handle, and has some charm.

Technically, there is nothing wrong with the camera. It has much of the same sensor and imaging pipeline components as their K-5 semi-pro camera model. It has sensor shift stabilization, otherwise only seen in the Olympus M4/3 camera models in this market segment. It can even accommodate old style screwdriver style focusing with legacy Pentax AF lenses.

Thus ends the era of the strangest mirrorless camera so far. It now sells at a reasonable price, and you have a chance to snatch up this piece of history if you are into collecting.