MattH

Would you mind posting a jpeg? I cant view that currently. Is that the 10-18 f4.5- 5.6 ? I doubt there will be any field curvature visible at that narrow an aperture anyway. I don't know if speed boosters reduce field curvature or maintain it. Perhaps they do maintain it. I have the 7.5mm which is why I'm very interested in the issue. It does have field curvature, but it that guys pocket4k was massively exaggerating the issue. The 20mm 1.7 has field curvature wide open too when focused to infinity so I'm worried that will look shit on the pocket4k also. It would be pointless getting one if all my mft lenses dont work properly.

No, I think its still there, it just seemed to die. I have been waiting for the update on that. Did he have a faulty camera? Did the problem go away once he stopped down further? I may need to bump it, because that seemed a serious issue.

Even the battery itself knows its about to die!

It's electronic rolling shutter so even if it had flash it would have a maximum flash sync speed of 1/50th of a second or something. Any faster than that and only part of the image would be exposing when the flash fired.

Ok. My second go. I forgot to look at the resolution of the large file. I guess I assumed it would have been scaled down. The horizontal resolution of 5212 is exactly the same as a M9 or M9-P, and there arent any other 18 megapixel full frame cameras i dont think. And the 18 megapixel apsc canons only have a horizontal resolution of 5184 so they are out. So the lens was produced in 1957 if its the M9 and 1955 if its the M9-P. I first thought it was 35mm but now I do think its 50mm. So 50mm summicron seems correct. It would have to be the 5cm screw mount colapsible summicron if it was 1955 so it is more likely An M9 with rigid 50mm Summicron as already guessed. So that was a nice run around the houses for no reason on my part. At least I got the exercise I guess. I presume it is wide open as the depth of field is shallow. If so, its certainly sharp enough wide open. I supose thats what this shows.

My guess is that the reason you are presenting this guessing game is because you were suprised in some way yourself. So Im guessing its not a new camera. So its an old digital, but it would probably have to have some significance in order for you to buy it. So like a 5d mark 1 or something. Could also be the first micro four third camera with a speed booster, but I'm getting a canon vibe from the image. It could be an old apc canon. Ah the 50d! We know you were already surprised by the fact that it can do raw video dispite not having a video mode. But because you already have that it would have to be a new lens or lens combination in order to suprise you. I initially thought it must be full frame. So I think you are surprised by how large sensor it looks. Im not a lens expert so someone else would have to work out what fast aperture lens it is. Unless you have removed the mount and mirror in order to cram a speed booster in. But thats probably pushing it a bit. As the 50d is 10 years old, 70 years total would put the lens date of birth at 1958 (the maths confused me for a minute). Searching 1958 lens brings up the mir 37mm f2.8, But the feild of view looks wider than 59mm equiv and the aperture looks wider too, so probably not, but then again if it was that I can imagine it surprising you.

Well that functionality is essentially built in to raw already. Your only choices are the physical exposure and the native iso. Even in prores, the way I understand it is that if you were at iso 100 and you adjusted your exposure so that the sky was just on the edge of clipping, you could increase the iso up to 1000 and the sky wouldn't clip. Just the average brightness would increase. So (not that I'm saying you said this) I don't think you would ever gain anything by shooting at one iso and then adjusting it so it resembled another. It would be preferable if you knew before hand. But if you are just talking about flexibility, I guess shooting in the middle of the range would make sense. Probably at the native iso's themselves of 400 or 3200. Then you would have a balance of noise performance and highlight latitude with a little bit of leeway either way. But to leave significant leeway to boost the brightness you would have to bake in what looked like an under exposed image and shadow detail would be lost due to insufficient bits in the shadows, or to leave significant leeway to crush the noise out you would have to bake in what looked like an over exposed image and highlights would potentially be clipped.

Perhaps I'm missing something, but why would you do that? (build a set of curves that turn the lesser tone curves into the good one). The way I saw it that with prores you are first choosing a native iso band and then the iso within that band defines were the midpoint is within the exposure range. With raw you just choose the native iso. The increments within those bands just alter how bright the image looks on your screen to act as a guide for how much light to put on the sensor. Are you talking about raw or prores?

From what I have seen I think the resolution plays the largest factor. The lower resolution of the og pocket helps to mask other amatuer/video looking things (lighting, colors, composition). A slight amount of added blur should soften up the image nicely and eliminate most of the difference between them. (For the record, this would be for narrative work only, if its a documentary you would leave it sharp. Though luckily this proviso is mostly redundant as most people sharpen anyway). I also think the og pocket may have had a tone curve that is favourable to highlights. The pocket 4k has different tone curves at different iso's (shown drastically by the dual native iso) so the discerning user will choose the ISO with a tone curve that matches the look they are going for at that particular moment (narrative/documentary), which is a trade off between highlight latitude and noize performance. In prores the tone curve is baked in, while the raw iso increments are only metadata but the native iso comes into the equation even in raw, and the chosen iso will indirectly effect how the user chooses to expose.

It will be very interesting if other companies adopt Blackmagic Raw. That will open things up massively.

Disregarding the arguments that go around about whether motion cadence is a thing, I think we can already tell how it will act with respect to motion: Its a rolling shutter, which seems to be fairly minimal due to the native 4k resolution. But it is there and it will be perceptible to some extent over a global shutter camera. But if you shoot raw there's nothing I can think of other than the rolling shutter that would effect motion. With either raw or prores you are getting rid of interframe compression which can effect motion in other cameras. And in raw you are getting rid of all compression (in case that would effect it)

Cool, well if they were happy with it, theres nothing anyone can say against it. The only thing the Pocket4k doesn't have compared to stills cameras is mechanical shutter and flash. In raw mode its more acurate to call it a 60fps electronic shutter stills camera with no buffer limitation than than it is a video camera.

Why did you choose to take stills with the pocket?

I summed up all that in a Laughing reaction. But thanks for spending the time to flesh it out.

Hmm. Maybe it cant scan partial lines? If the scan duration was proportional to the imaging area then yes it would be able to do 120p at a larger crop than 2x and be able to do 240p at 2x. So Im thinking the scan speed must be proportional only to the number of lines being scanned.

Yes! In fact Ettr literally means not blowing the highlights. It means exposing as far to the right on the histogram as you can without blowing highlights. (not counting things you choose to blow). It's only really aplicable in raw though.

The GH5s 120p is clearly line skipped. Evidenced by the jaggies. That's how it avoids cropping. A crop at 120fps when you are getting smooth details is really spoilt rich man problems.

It doesn't matter. Its half the price. My comment is simply refuting the claim that the 120fps on the GH5s is better. They both seem fine to me for what 120 fps is worth.

You are right. Looking at that Budapest clip the Pocket4k has more detail, but its smooth detail. Look at the straight lines on the plynth of the column. The GH5s has jaggies on straight lines. It's false detail that is giving the overal impression of sharpness. So this clip actually shows that the pocket is BETTER in 120 fps. The only flaw is the moire on that guys shirt that is when zoomed in which is actually there because of the extra detail and is only really visible zoomed in.

The crop is defined by the scan speed of the sensor. 120 fps means a frame every 8.33 milliseconds. If the entire sensor cant be scanned in that amount of time then full sensor readout 120 fps is physically impossible. The two options then are either to line skip the whole sensor or to do a full readout of a smaller proportion of the sensor. The pocket4k has clearly opted for the latter. I doubt the GH5 or GH5s have readout speeds of less than 8.33 milliseconds so if they have full sensor 120 fps then it is probably line skipped. I cannot, for the life of me, think why someone would need a 120 fps slow motion shot of a still detailed wide city-scape like that in which 60fps wasn't adequate. If you are doing 120 fps slow motion it is usually of something that is moving and thus detail is less important. If detail is really important then simply play 60 fps at half speed. In sports broadcasts they often achieve the slow motion by playing each frame about 3 times, and that looks smooth enough. The original pocket didn't even have 60fps. To me it just seems a silly ask to expect 4k detail 120fps full mft sensor from a camera of this price.

Light as opposed to heavy. Meaning small or minor.

ISO 1250 is the WORST iso to choose to test the red clipping issue. It is pulled 3200 iso, so even properly exposed it only has 2.3 stops of latitude above middle grey. So of course the highlights will clip and look shitty. ISO 1250 will be the number one noob trap on this camera. Im more worried about the wide angle lens blurryness. Yes, lenses have field curvature that is more aparent wide open and yes the Pocket 4k has a wider field of view that shows more of it, but that alone doesn't account for what the examples in that thread show. It must be the sensor stack thickness also. If stopping down to 2.8 doesnt resolve that issue then that is going to be a way bigger deal breaker than red clipping.

Ah ,that might be what I was thinking of. Im think I remember a review mentioning something about the the crop that was unexpected so that might have been it. I guess it is easier to get the 1080 by downscaling the 4k than line skiping a marginally larger area. Although that would lead me to think that perhaps the HD in crop mode has better detail than the line skipped full fram HD. Still I suppose the test is worth it just to see as really theres nothing physical to stop them choosing any crop they wanted.

The banding is fixed patten noise and happens on all canon sensors I think so its going nowhere. That's the major advantage with sony sensors which dont have this flaw. About the crop. Are you absolutely sure its cropping to 1.6 in apsc mode? might it be cropping more, therfore making the transition to 1.8 crop less? It would be easy to test with a tripod and another canon apsc camera with the same prime lens.

It would be good to see the same situation in prores. Shooting raw actually bypasses the majority of what the camera can do to fix the image, so criticising a raw image is really just criticising only the sensor, not the cameras handling of it. (which is another reason why blackmagic raw should be great as it could include image fixes). Specifically, try prores film at 1000 iso or 6400 iso exposed properly and then try reducing the exposure if the effect is still there. I wonder if a filter to lower only the reds would work? A green filter? And then adjust back the colors. Or maybe a diffusion filter.