KnightsFan

@blondini No, perspective distortion is only affected by distance from the camera to the subject. No matter what sensor/lens combination you use, if the camera and the subjects don't move, then the ratio of the size of two subjects will remain the same. I did a quick and dirty test to illustrate. It's a little imprecise (the camcorder would NOT focus on the guy in front...)
For all three images the camera is in the same place. I suspect the small discrepancies in ratio (2.2% error) are mainly due to moving parts inside the camcorder when it zooms, which changes its actual distance from the subject. But this is an easy thing to test yourself.
First image is a 4mm lens on a 1/4 type sensor

Second image is a 55mm lens on an APS-C sensor

As you can see, the ratio of the figures is the same. You could even use a wider lens and the ratio remains, because the distance has not changed:
Third image is a 2mm lens on the 1/4 type sensor.

 
Quote from Wikipedia:
 
 
 
Yes, it would. As long as the camera is in the same place, the relative size of the plane compared to the people will remain the same regardless of the lens or sensor. If you don't believe me or my Legos, go try it yourself!

Assuming we're talking about the same video, he was saying that manufacturers were erroneously marketing their lenses giving equivalent focal lengths, but not equivalent apertures. Like marketing an 14-42 f2.8 as a 28-84 f2.8. His point was that IF you change one number, you have to change the other. And he did specify that he was talking about equivalent depth of field - his title card for talking about it is "Aperture & Depth of Field." He was using ISO to talk about exposure, which he explained early in the video. In his logic, once you compensate for exposure with ISO, you have to compensate with aperture as well.
It is a roundabout logic, but it does account for the lower light gathering power of a smaller surface area, and thus the need for more gain (and thus more noise or lower resolution) to reach the same ISO.
 
True, but the light gathering power of the space is directly related to its size. The SNR ratio of a smaller sensor will be lower than a larger sensor (assuming similar tech), given the same image scaled down. Hence the f0.95 doesn't actually have any low light advantage over the f2.0, if you look at the system as a whole.
 

@Andrew Reid Yeah, "directly related" isn't the correct wording, but we mean the same thing. As sensor size increases, you can either increase resolution with the same pixel size, or increase pixel size at the same resolution. You said so yourself in your disclaimer at the end: "For example you can have higher megapixel counts because there’s simply more real-estate on the chip surface to add more pixels."
That is why I said "assuming similar tech." The a7r3 has better tech. I should have specified photo-sensitive area, rather than simply "surface area" and would perhaps have been clearer.
 
Wait, you just posted an  article about how the 17.5mm f0.95 is comparable to the 35mm f2.0. If you don't have to change both numbers, why aren't you saying the 17.5mm is comparable to a 35mm f0.95?
But anyway, I'm just explaining Northup's argument. He claims there is marketing material on their websites that is misleading, idk if that's true but his crop factor equivalence works for the comparison he's making, despite being poorly worded and convoluted.

@blondini No, perspective distortion is only affected by distance from the camera to the subject. No matter what sensor/lens combination you use, if the camera and the subjects don't move, then the ratio of the size of two subjects will remain the same. I did a quick and dirty test to illustrate. It's a little imprecise (the camcorder would NOT focus on the guy in front...)
For all three images the camera is in the same place. I suspect the small discrepancies in ratio (2.2% error) are mainly due to moving parts inside the camcorder when it zooms, which changes its actual distance from the subject. But this is an easy thing to test yourself.
First image is a 4mm lens on a 1/4 type sensor

Second image is a 55mm lens on an APS-C sensor

As you can see, the ratio of the figures is the same. You could even use a wider lens and the ratio remains, because the distance has not changed:
Third image is a 2mm lens on the 1/4 type sensor.

 
Quote from Wikipedia:
 
 
 
Yes, it would. As long as the camera is in the same place, the relative size of the plane compared to the people will remain the same regardless of the lens or sensor. If you don't believe me or my Legos, go try it yourself!

@Andrew Reid Yeah, "directly related" isn't the correct wording, but we mean the same thing. As sensor size increases, you can either increase resolution with the same pixel size, or increase pixel size at the same resolution. You said so yourself in your disclaimer at the end: "For example you can have higher megapixel counts because there’s simply more real-estate on the chip surface to add more pixels."
That is why I said "assuming similar tech." The a7r3 has better tech. I should have specified photo-sensitive area, rather than simply "surface area" and would perhaps have been clearer.
 
Wait, you just posted an  article about how the 17.5mm f0.95 is comparable to the 35mm f2.0. If you don't have to change both numbers, why aren't you saying the 17.5mm is comparable to a 35mm f0.95?
But anyway, I'm just explaining Northup's argument. He claims there is marketing material on their websites that is misleading, idk if that's true but his crop factor equivalence works for the comparison he's making, despite being poorly worded and convoluted.

Oblique rays strike the center of the image as well as the corners.  These are aperture dependent and are called marginal rays.  The relationship between f-number and the marginal ray angle is given by f/# = sin(theta), where theta is the marginal ray angle.  So, at f/1 the marginal ray angle is 30 degrees, and so on.
The obliquity of the marginal ray does cause a form of pixel vignetting that perhaps should be called pixel apodization.  The effect is that the edges of bokeh blurs - where the marginal ray obliquity is highest - is darkened relative to the center.  The good news is that this is the definition of "good bokeh".
I think Andrew's two sample images reveal this effect.  I would guess that the bottom image was shot with the faster lens.
 

Companies would love it if everyone believed this , but unfortunately (or fortunately for our wallet) it is not the case.
I have built and bought too many computers over the years.  Currently I am using a MacPro from 2010 with 12 cores, 64gb ram, an M.2 NVMe boot drive and a Titan X GPU and I can tell you this is no way obsolete:
 
That was of course when Apple made computers for professionals  . But with windows, which have even more choices, you can easily update any part of the system. Even with different generation CPUs you just need to upgrade the CPU+motherboard. RAM technology does not change that fast and easily lasts for ~10years. 
Future-proofing is very important when building a computer. Unless someone has either too much cash to waste or no idea how to build a computer that is. 

@blondini No, perspective distortion is only affected by distance from the camera to the subject. No matter what sensor/lens combination you use, if the camera and the subjects don't move, then the ratio of the size of two subjects will remain the same. I did a quick and dirty test to illustrate. It's a little imprecise (the camcorder would NOT focus on the guy in front...)
For all three images the camera is in the same place. I suspect the small discrepancies in ratio (2.2% error) are mainly due to moving parts inside the camcorder when it zooms, which changes its actual distance from the subject. But this is an easy thing to test yourself.
First image is a 4mm lens on a 1/4 type sensor

Second image is a 55mm lens on an APS-C sensor

As you can see, the ratio of the figures is the same. You could even use a wider lens and the ratio remains, because the distance has not changed:
Third image is a 2mm lens on the 1/4 type sensor.

 
Quote from Wikipedia:
 
 
 
Yes, it would. As long as the camera is in the same place, the relative size of the plane compared to the people will remain the same regardless of the lens or sensor. If you don't believe me or my Legos, go try it yourself!

@blondini No, perspective distortion is only affected by distance from the camera to the subject. No matter what sensor/lens combination you use, if the camera and the subjects don't move, then the ratio of the size of two subjects will remain the same. I did a quick and dirty test to illustrate. It's a little imprecise (the camcorder would NOT focus on the guy in front...)
For all three images the camera is in the same place. I suspect the small discrepancies in ratio (2.2% error) are mainly due to moving parts inside the camcorder when it zooms, which changes its actual distance from the subject. But this is an easy thing to test yourself.
First image is a 4mm lens on a 1/4 type sensor

Second image is a 55mm lens on an APS-C sensor

As you can see, the ratio of the figures is the same. You could even use a wider lens and the ratio remains, because the distance has not changed:
Third image is a 2mm lens on the 1/4 type sensor.

 
Quote from Wikipedia:
 
 
 
Yes, it would. As long as the camera is in the same place, the relative size of the plane compared to the people will remain the same regardless of the lens or sensor. If you don't believe me or my Legos, go try it yourself!

I would get a MoBo with more PCI slots. They can be used for many expansions, including GPU, I/O ports, capture cards, WiFi cards, and more.
I would also get a case with lots of large fans. I've got something like 3 120mm and 2 140mm fans, which keeps it cool enough to prevent the loud GPU fans from kicking in. Lots of large fans => low RPM => less overall noise.

@horshack If I'm not mistaken, pixel vignetting is entirely due to angle of incidence on the sensor. Smaller sensors have less oblique angles of incidence at the corners. I could be wrong, but I bet that pixel vignetting won't be a factor in this comparison. I'd love to see evidence either way, though.

@horshack If I'm not mistaken, pixel vignetting is entirely due to angle of incidence on the sensor. Smaller sensors have less oblique angles of incidence at the corners. I could be wrong, but I bet that pixel vignetting won't be a factor in this comparison. I'd love to see evidence either way, though.

Interesting article and great job matching the focal lengths and depths of field, but the cost associated with getting the full frame look with the GH5 will put you at $2500-$3000. With an a7iii and a Nikkor 35mm f/2, you can be well on your way for less than $2200, plus you’ll have the option to go with f/1.4 and 1.2 lenses and higher ISOs.
The Voigts are a great cheat though and the 25mm is one of the best lenses I’ve ever used. 
 

Hi!
An update.
Now I ordered lots of stuff (and waiting for it to arrive)
This is what I got (incl freightcosts):

5DII with battery grip - 660€
- not the cheapest one but seemed to be well treated and in good condition, got it down from 800€, 17 000 shutter count
Three Nikkor ais lenses (and a Nikon f301) 24mm f/2.8, 55mm micro f/2.0, 85mm f/2.8  - 310€
One ritzgear 52mm ND-filter 36€
One Komputerbay 128GB 1066x cf-card  - 103€
three lens adapters (no brand china) ~25€
One SanDisk ImageMate Pro USB 3.0 Multi-Card Reader SDDR-489 - 27€
Total about 1161€

Figured to get an external hard drive aswell, but I think I will start off with the storage I got at the moment. The 5DII got a little bit more expensive than I had thought to pay but otherwise I'm quite pleased with what Ive got within this budget. fingers crossed everything will work alright. 

Hope all is well with you people
/ Erik

I agree with KnightsFan, I’d go with a 50mm, either a 1.8 or a 1.4 and decide from there if you prefer a wider or longer lens. When I used to shoot on crop sensor cameras, I liked to be between 35 and 50mm after the crop, so I never owned too many 50mm lenses. Now that I have a FF camera, I am discovering and testing a bunch of lenses, I never would have bought before. And it’s great because 50mm lenses are cheap.
The EF Mount is so adaptable that there are a bunch of vintage lenses that could work. Nikkors are great because they are known for their cinematic look at a good price point. But if you are only looking for a few manual primes, there are also Contax Zeiss or Olympus OM or Yashica ML or Pentax that also have some great choices. And since the advent of the cropped sensor, there are so many cheap 50mm lenses on the used market to experiment with. 
Now I have purchased 99% of my lenses from eBay. As long as you buy from a seller with a good feedback rating and a detailed post, you should be fine 90% of the time.
Just don’t do what I did when I started and buy based on price, you’ll end up with a lot of junk that is hard to resell and collects dust. So know what you’re looking for in focal length and lens speed...
I shoot run and gun, horror and thriller short films. So I need fast lenses for lowlight. I try not to get any prime lens slower than an f/2. But I also rarely change lenses, so a fast 50mm lens (which are common) is often all I’ll use on a day’s shoot. Your mileage may vary. 

Your lens choices seem reasonable to me. You could grab a 50 to start with, and see whether you find yourself wishing for a longer or shorter lens. I personally would go for manual focus primes. I wouldn't say that it matters all that much in terms of how easy it is to start with.
I still don't have an ND filter. It's only crucial if your camera has a high base ISO, like 800 or 1600 on some Sony cameras. You should get a polar filter though, which can more or less be used as an ND filter as well.
I have a Nikon->Canon adapter for each lens. They're cheap, and a pain to get on and off.
Yeah, a new speed booster costs almost as much as a used 5d2, and there's no mirrorless APS-C cameras that shoot raw. So it doesn't seem like it's a good fit for you.
You'll be able to get a repeatable process with raw, though that sort of defeats the purpose of having all that flexibility. But like you say, it's always nice to have if you need it for a specific project.

@mercer Not sure if you're referring to me, but I'm not opposed to Magic Lantern. I absolutely love it!
But I do want to be honest about workflow challenges so deatrier can make an informed decision. Raw, and especially Magic Lanter, is certainly most appealing to those of us who love tinkering, and have plenty of hard drives. Some people might not be into that.

Based on your interest in full frame Raw, then really it's either the 5d2 or stretch and get the 5d3 if you can find an amazing deal.
Obviously, many of us have different priorities, but if that's what's important to you, then it's basically down to those two and seems to be mainly an economic choice.
 
Electronic viewfinders and screens are actually easier to focus with, because you can usually do a pixel-to-pixel magnification and/or focus peaking. Optical viewfinders can be more fun to use, but focusing with an evf is easy, too.
Shooting Raw on a Canon camera is not simple. The post workflow isn't bad, though it will take some experimentation. The main problem is that when shooting HD raw, you get like 5 minutes from a 32GB card. Compressed raw and lower bit depths were announced since the last time I used ML, so the situation is a little better now, but it's still very different from shooting all day on a 32GB SD card like you can with H.264. If you go raw for an all-day shoot, you will either need a LOT of cards, or you will need to dump cards to a hard drive as they fill up.
Not to dissuade you. Like I said, the image is worth it. But you will need to budget for cards and hard drives.
I have a set of Nikkor K lenses that I love. Great character (not terribly sharp, lots of distortion and CA, but wonderful vibrant colors), great handling, they can adapt to anything, easy to declick (if you're into that). Would definitely recommend them for the price.

What kinds of projects will you be doing with it: Narrative or documentary? Studio or run'n'gun? Will you be filming in low light? Do you plan to do a lot of color grading? Green screen or VFX? Do you also want to take photos?
If you're set on ff, I don't really know of any options that would be better in the price range you've set. You can make great stuff on a 5d2--or any other camera from the past ~6 years. If it were me, though, I'd get a used APS-C camera, especially if you already plan to upgrade in a year or two. You could get a cheaper Canon for a few hundred dollars, or, if you can find a good deal, get an NX1. It's a fantastic all-rounder. I got mine used a year ago for $850 (700€) with lens and accessories included.
If you plan on using Magic Lantern for raw video, be aware that it will be a mild PITA. Totally worth it, but not easy.
Do you have an external microphone or any other audio gear? I'm not a fan of the H4 myself.

That is not the point at all, and it's half the price of what you are posting anyway, 50% less,  if people believed that a device such this can replace a proper monitor, then he/she was mistaken and I hope they didn't support the kickstarter campaign, because they are in for a big disappointment.
I quote myself "I see it as a backup option, for whatever reason, maybe put a tablet as a directors monitor, maybe you do not have another monitor, maybe for a second camera monitor, maybe you forget your monitor(!) or its batteries, and so on and so forth, and my mobile phone is 2K 600nits IPS 5.7", so I would like to see how it goes!"
and the next phone, will have definitely a better screen, and I hope I still be able to use it with the Lukilink.
I never saw it as a proper pro field monitor replacement, and I have yet to see any of the Feelworlds as one.
(this Feelworld says 460 nits, not even close to a mobile phone, and probably the real value is less than that, and nits are very important, because the first thing you expect from a monitor, is to be seen).

The mewlips controller is unusably laggy for me also.
After finding that the remote viewfinder app only worked with native lenses and the mewlips was unusable, I too searched for a way to monitor with a smartphone. I found nothing that was inexpensive. Apparently, you can use some sort of USB capture card and OTG cable for Android, but it would have been expensive and inelegant.
Hopefully, the LukiLink project will finish soon because that seems like the solution you and I are hoping for.
If you're on a tight budget and just need an external screen, you can get Raspberry Pi monitors for cheap.

The mewlips controller is unusably laggy for me also.
After finding that the remote viewfinder app only worked with native lenses and the mewlips was unusable, I too searched for a way to monitor with a smartphone. I found nothing that was inexpensive. Apparently, you can use some sort of USB capture card and OTG cable for Android, but it would have been expensive and inelegant.
Hopefully, the LukiLink project will finish soon because that seems like the solution you and I are hoping for.
If you're on a tight budget and just need an external screen, you can get Raspberry Pi monitors for cheap.

@Mokara I read your entire post. It sounded like you were saying HDMI video by nature is debayered and processed. Even the second part of your post seemed to imply that Atomos was just sort of hoping that someone could send a Raw signal for them to record. Sorry if I misunderstood.
Raw photos aren't processed, so there is a way for the image to bypass the hardware processor. And besides, Atomos seems to think it can be done.
Edit: If Magic Lantern can write Raw to disk using software only, I have no doubt a firmware update to output Raw over HDMI is technically possible for many other camera models.

@Mokara I read your entire post. It sounded like you were saying HDMI video by nature is debayered and processed. Even the second part of your post seemed to imply that Atomos was just sort of hoping that someone could send a Raw signal for them to record. Sorry if I misunderstood.
Raw photos aren't processed, so there is a way for the image to bypass the hardware processor. And besides, Atomos seems to think it can be done.
Edit: If Magic Lantern can write Raw to disk using software only, I have no doubt a firmware update to output Raw over HDMI is technically possible for many other camera models.

No need to feel sorry for us PC users. Resolve has been cutting through 4K raw like butter for years.
I've been shooting raw exclusively since 2013. Stopped using proxies in 2015. I've only ever used regular consumer hardware for post. Frankly, raw is old news for PC users.