John Matthews

I can try to upwrap this portion of the video for you. Regardless of how much compression was used, his point was to show that, even with a 4k or 8k image, compression plays a significant role in the final image. Just increasing the megapixels isn't enough and would decrease the quality, not increase it. Concretely, a 4k image compressed to 10mbps will not produce a better image than a 1080p image at 10mbps. His point was to show resolution doesn't necessarily mean a quality image; there are many factors, compression being one of them. IMO, the issue of Long GOP and ALL-I is mute with modern encoders, but, if looking for a formula, 50mbps long-gop is equal to 100mbps ALL-I. You'll get significant space savings and 99% the quality in most situations. I cannot comment on working with Raw... I'd never do that for my purposes.

Concerning my previous post, I do think that Rodney Charters and Steve Yedlin differ in their theories about acquisition of the image. It seems Charters looks for the camera that gives the best image where as Yedlin prefers the most neutral image as possible, then add character in the post pipeline.

I remember watching both of these. In a nutshell, up-resing tech was good enough for them in 2014, I imagine it's a little better in 2021. They'd much rather work like that and continue with a speedy 2k pipeline. And once again, audiences cannot tell the difference.

For me, he effectively demonstrates the insignificance of taking professionally prepared 4k+ content, downscaling it to 2k, and upscaling it to 4k again. The resulting images, even when compared A/B style, don't show any difference. I'd love for you to prove otherwise. I really didn't think of it like this until after watching him. Again, his point wasn't necessarily this though- it was to show there are many other considerations BEFORE pixel count that show significant importance as long as the detail threshold is met.

As almost no one consumes imagery in print form, the issue of megapixels is completely mute nowadays. Any camera is good enough and will satisfy any non-forensic based analysis. Your A7s should do a splendid job. You just can't crop the hell out of it without AI upresing it if you don't think it'll satisfy people. As I remember it, the only thing you need to "worry" about with that camera is highlight fidelity in jpeg's and video with its tendency to go cyan.

After watching the video, I was surprised how many localized edits were made, incredible amounts of finessing the image. The director doesn't like pink; so, we choose a color-space with less pink, but still has significant color saturation and separation. Of course, they shoot it HDR too, but Fincher doesn't that kind of look. "Arduous" was the right word. However, when I watch the Steve Yedlin video, I really don't get the same impression. However, not knowing the entire process, it might be the same. The video you linked makes me never want to be a colorist!

I'd recommend you watch the whole video. It was rather eye-opening for me. His point is to gather data without any imperfections if possible and add value to his content through a streamlined image processing pipeline, regardless the camera used to capture. I highly doubt any viewer would ever see a flaw with his strategy. I'm aware that sharpness is not detail... and he covers that in the video too. Another major point is that no manufacturer is making a new human retina; therefore, the maximum detail has already been hit (even with 1080p!). Any more efforts at showing more detail would require sitting much closer to the content at which point you'd find yourself moving your head around to see the scene, taking you pointlessly out of the story. No viewer wants that.

I'm fairly certain he primarily shoots with the Alexa in 3k as, according to him, it gives the best image regardless of being "only" 3k. When you look at the way he presents, it seems clear that he's right. In the video, he talks mainly about using a camera as a data collection device and wants ALL sensor/film imperfections to be absent if possible. This way, he can add the imperfections in post to give the look he wants (grain, color, halation, etc.) through a comprehensive editing pipeline. Resolution was the ONE criteria that he didn't need as it met the accepted threshold of detail, even in a camera comparison situation, let alone hidden within a story.

Yes, as he says in the video, people are just looking at that ONE number to make easy choice as to which camera is better. Maybe this is what separates a real cinematographer from wannabes. The image is what counts, not the megapixels (after you get to the "accepted" amount of detail threshold).

Sorry, typo in title. His last name's "Yedlin", not "Yeldin". Sorry to Steve. Too late to update.

I've been watching some resolution insights by cinematographer Steve Yeldin that I think many might find very interesting. Not sure if this has already been posted... It would be interesting to discuss.

My suggestion would be to never buy anything new- you won't lose as much. Just be patient and wait for a good deal to come up. In the end, you'll find you'll lose little to no money on the "investment". Be the guy who gets the awsome deal, not the guy who loses all that money. Just an idea.

I found a PL9 for 350 euros with a 14-42 ez pancake, and a 45 f1.8. You'll be well under 200 euros if you resell the lenses. Here's the listing:

Let's just keep it going... sell your m43 stuff NOW! You don't want to lose all that money you put into your "investment". We all know that those once great images that everyone liked in the past will only be considered garbage in the future. It's all about the camera.

I've had similar luck finding some good deals, but I've also seen price increases in Olympus gear recently ...no idea why. Now, Panasonic stuff is super-inexpensive. The only exception is the e-m1 ii, which is still a super-killer deal right now. I just love those guys who keep spewing "sensor size is everything". Let's keep that going for as long as possible!

My Olympus PL9 was just under 200 Euros (body only). It has the same IQ as the e-m10 iii, but it doesn't have an EVF or 5-axis IBIS (it only has 3, but it's excellent).

I got a gh2 for about 100 euros... not 4k, but it does have 1.86 crop. For my usage, it makes sense. I hacked it and it has an interesting look to it... different from other Panasonics I've used in the past. I still find 4k slow to edit.

So Sigma's promoting this lens as a "family" lens? I'm sure it would do fine, but this system seems a little over-kill. I'd buy if I had an fp though.

Just an idea... why not use manual focus and let the person just go slightly in and out of focus? ...much less distracting than electronic pulsing.

So, I got a little-used (9500 shutter count) GH2 today for reasonably cheap. I'm actually amazed at all the features and buttons it has. For 2010, that's great! My question is : what's best hack for it? If I'm not mistaken, Cluster-x was the most recent? I'm more inclined to use the longGOP versions (Drewnet), something that doesn't fill up my disks but offers very decent 1080p 25fps. I'd love some input! Also, I'd like to use it for live-streaming. I already have a Blackmagic Ultrasudio recorder and I'm 99% sure it's fully compatable.

Go to video -> Sound -> audio setting -> PCM Recorder Link -> Camera REC volume. You need to have it set to "disabled". Now there's no more hiss, but you need to have your external recorder connected to the mic input as it is now the preamp and the camera does nothing (except record the audio it receives). Took me forever to figure this out. I thought someone might be interested.

Apple's going blow everyone away in the short term. Their willingness to let go of old tech (for better or worse) and prioritize the performance/energy efficiency factor give them a serious edge. The icing on the cake is the ability to match software and hardware, fine-tuning performance for better results.

That's brilliant!

Yes. A great song... just very surprised at the video.

The pitch: We're going to make the most epic music video ever! There are going to be torch jugglers in slow-mo while singing, a flashback in a query involving child look-alikes in bathrobes at a make-up table, motorcycles in a parking garage filled with water smoke while playing instruments, Daft-punk style helmets in front of the drum set, singing while moving on the motorcycle, and as a grand finale, a bulldozer knocking over a building and then torching it with a flamethrower. With that kind of production value, what could go wrong? Take a look: