A7s 5D MK3 Raw shootout
- Thread starter squig
- Start date
Samuel- I have seen yellow splotches in GH4 images as well (almost ruined a great still shot). It appears to be a form of medium to very-low frequency noise, perhaps from higher frequency noise which gets clumped together into lower frequency noise during in-camera noise reduction. I've seen red/magenta splotches in the 5D3 which seems to be less of an issue with skintones.
eddavid- while anything is possible it's not likely the a7s will be hacked. The Canons were a special case and the a7s is pretty good as is. We would expect the next FF Canon to leapfrog the a7s with no hacks.
eddavid- while anything is possible it's not likely the a7s will be hacked. The Canons were a special case and the a7s is pretty good as is. We would expect the next FF Canon to leapfrog the a7s with no hacks.
dfreerider
Well-known member
Well, anything is possible, and I hope it happens, but the Magic Lantern guys have been at it for many years. A group of Nikon hackers have been working on the Nikon DSLRs for a couple of years ever since one of them managed to decrypt the firmware, and while they are making progress, they are still very far away from magic lantern type functionality.
Personally I think the A7s will be superseded by newer models offering 10 bit recording, long before a raw hack is feasible. But I'd love to be proven wrong
Personally I think the A7s will be superseded by newer models offering 10 bit recording, long before a raw hack is feasible. But I'd love to be proven wrong
Well, anything is possible, and I hope it happens, but the Magic Lantern guys have been at it for many years. A group of Nikon hackers have been working on the Nikon DSLRs for a couple of years ever since one of them managed to decrypt the firmware, and while they are making progress, they are still very far away from magic lantern type functionality.
Personally I think the A7s will be superseded by newer models offering 10 bit recording, long before a raw hack is feasible. But I'd love to be proven wrong
I think you're right.
squig
Veteran
Yeah that yellow thing is weird, looks like CA down the right side of the mic. Definitely worse with prores, so much so that I've abandoned prores recording until the shogun comes out. I thought it was a lens or an adapter doing it but I've tried different combinations and it's still there. Changing in camera saturation doesn't help either. It's making the image softer than the MK3 raw. It's not occurring with raw stills.
A7s raw still
squig
Veteran
Theoretically sure. Just a lot of time & effort to reverse engineer the code.
If it's true that it's capturing a 2.7k crop off the sensor it's perfect for a raw 2k recording.
Last edited:
Samuel H
Major Contributor
* I doubt we'll get a RAW hack for the a7s. It would be great (that RAW still is a lot better than both internal and external video), but there was a NEX hack more than a year ago and it went nowhere (they only managed to allow new languages to be installed).
* I doubt the next Canon will have better DR in RAW stills than their current models. At some point they'll change the architecture and leap forward to match Sony, Nikon, etc, but the optimal forecast is always saying it won't happen this time, and being wrong only once: when it happens.
* I doubt the next Canon will have better DR in RAW stills than their current models. At some point they'll change the architecture and leap forward to match Sony, Nikon, etc, but the optimal forecast is always saying it won't happen this time, and being wrong only once: when it happens.
I totally agree with you. The problem is that Canon is trying to protect their C-line. That means - I think - that they will not do as good "video camera" as the A7s. Maybe later they will realize they are losing the market, but it will for sure take some time. However in that case they will need to reconsider their whole lineup and prices...I do not count on them right now...
Back to the topic. The 5d3 files look much better. I thought that the yellow cast shows only at shadows, but I was very-very surprised that even in highlights at window there was yellow color cast. I thought it is probably some sort of noise, but this example shows that it does not affect only the dark parts of the image. What a shame that we can not find a solution to remove it straight out the camera.
squig
Veteran
After 3 weeks of testing I think the A7s is best suited as a b-cam for the 5D MK3 raw. The biggest issue is the 8bit S-log files (including prores), there's not much info there and it's a lot harder to get a solid image in post (this may change when new LUTs come out). The A7s is very good in high contrast scenes (as long as you can avoid the highlight aliasing) and it does have a bit more dynamic range than the MK3. Beyond 3200 ISO the A7s has the edge but for everything else I think the MK3 is a better camera. There's nothing about the A7s that makes it worth giving up the versatility and quality of a digital negative which you can manipulate as much as you want in post. There's just too many compromises with the A7s, it's just not good enough to give up raw and go backwards. I hope somebody does a raw hack on the A7s but until then I'll just use it for production stills and for low light scenes. Having said that I've been known to change my mind on occasion
Very interesting and helpful thread, thanks. I was very very close to selling my mk3 in exchange for an a7s but now I'm not so sure. The huge file sizes for ML RAW and the relatively low quality output of its internal H.264 recording make it a hard one for me to fully embrace in most production environments.
Do all the benefits of the a7s in terms of latitude and low-light performance vanish once you move out of S-LOG? Because I would be tempted if one of the more baked-in modes produced better color and still preserved most of the DR at the cost of having contrast more baked in.
Do all the benefits of the a7s in terms of latitude and low-light performance vanish once you move out of S-LOG? Because I would be tempted if one of the more baked-in modes produced better color and still preserved most of the DR at the cost of having contrast more baked in.
squig
Veteran
You lose all the extra DR with the baked in profiles, shooting like that just gives you a regular DSLR H.264 look but with more resolution. You still have the latitude after grading S-log but because you can't manipulate the image like you can with raw it kinda evens out. The only way I'd want to shoot with the A7s is 4k recorded to a Shogun with a 1080p downscale and 10bit luma conversion but then it's still more work in post and bigger files than the MK3 raw.
Samuel H
Major Contributor
After 3 weeks of testing I think the A7s is best suited as a b-cam for the 5D MK3 raw. The biggest issue is the 8bit S-log files (including prores), there's not much info there and it's a lot harder to get a solid image in post (this may change when new LUTs come out). The A7s is very good in high contrast scenes (as long as you can avoid the highlight aliasing) and it does have a bit more dynamic range than the MK3. Beyond 3200 ISO the A7s has the edge but for everything else I think the MK3 is a better camera. There's nothing about the A7s that makes it worth giving up the versatility and quality of a digital negative which you can manipulate as much as you want in post. There's just too many compromises with the A7s, it's just not good enough to give up raw and go backwards. I hope somebody does a raw hack on the A7s but until then I'll just use it for production stills and for low light scenes. Having said that I've been known to change my mind on occasion
Well, the good thing about living in the future is that you have some great options at great prices. If you can afford to have both, you don't need any of them to be the perfect camera.
BlackLegSanji
Member
But in the sample footage thread I saw you confirm Mcgregor's finding that Slog2 and Cine4 has the same amount of usuable DR. Can you explain what's different here?
cyork
Well-known member
Cine4 is great but I wouldn't shoot it thinking I was getting the same amount of usable DR as Slog2. Certainly won't net you the same image with the same amount of midtone snap. Not by a long shot. Tests are interesting but I wouldn't take anything as gospel until you get your own hands on the camera and shoot the footage side by side yourself THEN apply your own workflow and taste. MMV
