Canon Has Solved No IBIS On EOS R

[Django]

 

2 minutes ago, liork said:

As you said, as long as the EIS has extra crop and softness, it cannot be "close or superior" to IBIS. On the day it will overcome these 2 problems, then I will prefer that.

 

I meant close or superior in terms of actual stabilization performance, not overall usability. I prefer IBIS too, especially for stills.

[DBounce]

7 minutes ago, liork said:

As you said, as long as the EIS has extra crop and softness, it cannot be "close or superior" to IBIS. On the day it will overcome these 2 problems, then I will prefer that.

Softness, which is virtually unperceivable is preferable to the jello effect and abrupt shifts that come with hardware stabilization. It's one of the reasons I happily ditched the GH5 for the S version.

I think IBIS for stills is fine... even desirable as unstabilized lenses benefit from it. 

[frontfocus]

14 minutes ago, Django said:

As explained above, EOS R's EIS communicates with the gyroscope as well as the lens IS to give you a result close if not superior to 5 axis IBIS.

I'm not seeing that. Again, the picture is captured and there is shake in it. The camera then corrects the image by manipulating it. 
With any hardware IS solution you try to compensate without manipulating image data. You can of course do electronic stabilization as addition, but as only form of stabilization it's insufficient. 

[liork]

2 minutes ago, DBounce said:

Softness, which is virtually unperceivable is preferable to the jello effect and abrupt shifts that come with hardware stabilization. It's one of the reasons I happily ditched the GH5 for the S version.

So you use Gimbal I suppose. Do you also add sometimes digital IS in post?

[DBounce]

4 minutes ago, liork said:

So you use Gimbal I suppose. Do you also add sometimes digital IS in post?

I use a Ronin S. But it's nothing I carry casually. I almost never post stabilize. If I want stable footage I will use a jib, slider, gimbal, or if I need repeatable movements, or working alone, I will use a full blown motion control system.

[Django]

9 minutes ago, frontfocus said:

I'm not seeing that. Again, the picture is captured and there is shake in it. The camera then corrects the image by manipulating it. 
With any hardware IS solution you try to compensate without manipulating image data. You can of course do electronic stabilization as addition, but as only form of stabilization it's insufficient. 

Again, by using 2-way communication data of the gyroscope and lens IS info, the in body EIS is capable of accurately compensating shake/movement instead of just guessing it through an algorithm in post warp stabilization.

[frontfocus]

3 minutes ago, Django said:

Again, by using 2-way communication data of the gyroscope and lens IS info, the in body EIS is capable of accurately compensating shake/movement instead of just guessing it through an algorithm in post warp stabilization.

the only thing you remove is the need to calculate/guess the movement before correcting, the process is the same. 
But I'd love to see how it works, if anybody with an Eos R and a 50mm f/1.2 or adapted non stabilized lens could upload a few videos. That would be nice. Handheld, slow walking, so we get different types of movement. 

[DBounce]

54 minutes ago, frontfocus said:

the only thing you remove is the need to calculate/guess the movement before correcting, the process is the same. 
But I'd love to see how it works, if anybody with an Eos R and a 50mm f/1.2 or adapted non stabilized lens could upload a few videos. That would be nice. Handheld, slow walking, so we get different types of movement. 

Plenty of hand held shots in this video. In the comments the videographer states he is confident of using it hand held. Also, I think the image looks fantastic. Certainly good enough to tell a story. This kind of work is what happens when you forget specs and just shoot.

 

[liork]

But you can say "image good enough to tell a story" about most of the cameras in market today... even on some smartphones.

You can also say "forget specs and just shoot" about any camera.

All these general sentences teach nothing when you try judge a camera.

[DBounce]

4 minutes ago, liork said:

But you can say "image good enough to tell a story" about most of the cameras in market today... even on some smartphones.

You can also say "forget specs and just shoot" about any camera.

All these general sentences teach nothing when you try judge a camera.

I guess what I'm saying is don't judge this camera by the specs alone. It's output and image quality is pretty dam good. Is it soft? Well in the default setting sure... but with most cameras the opposite is true... and what do we do? We almost immediately dial down sharpness. With the EOS R you just dial it up to level 2 and you are done. I'll often try to decrease sharpness by adding filters like the BPM to my lenses. 

By "forget the specs and shoot", I mean this camera doesn't get in your way. Ease of use and convenance go a long way to make one more productive.

[Snowbro]

Not an eos r fan really, but I saw a detailed video (dont remember where) & the guy showed that you need to do at least +3 sharpness in clog 4k. Otherwise it is so soft, you cant fix in post. Something odd happening there. 

[Mokara]

6 hours ago, liork said:

As you said, as long as the EIS has extra crop and softness, it cannot be "close or superior" to IBIS. On the day it will overcome these 2 problems, then I will prefer that.

As long as you have minimal rolling shutter and an oversampled sensor, using EIS should be superior to IBIS.

The weird distortions you sometimes get with IBIS are likely due to the sensor rolling in the focal plane. You get similar effects with lens stabilization, which also results in the sensor rolling in the focal plane relative to the lens elements. How severe those artefacts are will depend on how aggressive the mechanical stabilization is.

[liork]

8 hours ago, Mokara said:

As long as you have minimal rolling shutter and an oversampled sensor, using EIS should be superior to IBIS.

Well, still waiting for the first Mirrorless camera to show it. Right now, all present models show sharpness decrease in EIS.

[UncleBobsPhotography]

4 hours ago, liork said:

Well, still waiting for the first Mirrorless camera to show it. Right now, all present models show sharpness decrease in EIS.

Fast shutter speeds is the only way to solve the softness issues with EIS. Maybe they'll be able to improve EIS with spatial oversampling in the future.

[Mokara]

13 hours ago, liork said:

Well, still waiting for the first Mirrorless camera to show it. Right now, all present models show sharpness decrease in EIS.

That is because most MILCs don't oversample enough or don't oversample at all, so some degradation happens.

The NX1 did a pretty decent job at it though.

[Snowbro]

On 1/2/2019 at 8:54 AM, frontfocus said:

the only thing you remove is the need to calculate/guess the movement before correcting, the process is the same. 
But I'd love to see how it works, if anybody with an Eos R and a 50mm f/1.2 or adapted non stabilized lens could upload a few videos. That would be nice. Handheld, slow walking, so we get different types of movement. 

I did this in 1080p 24p with a sigma 35mm 1.4 @ f2. It wasnt terrible until low light, then I saw lines in the sidewalk jumping around. 

Let's be honest, canon has 720p upscaled to 1080p? Lol I dont want to reduce that res any more than it is. What is the eos r 120p 720p? Upscale 480p? Haha. In 2019, I really dont care to shoot in 1080p now, it's so mushy in outdoor scenes. I think it's fine for a talking head vlog sequence, but not much else. Most people probably wont even know if a quick talking head shot on an airplane, was filmed on a phone or an m50 etc. 

[liork]

9 hours ago, Mokara said:

That is because most MILCs don't oversample enough or don't oversample at all, so some degradation happens.

The NX1 did a pretty decent job at it though.

Well, some MILC oversample from 6K and there is still degradation in EIS.

[Mokara]

On 1/3/2019 at 10:34 PM, liork said:

Well, some MILC oversample from 6K and there is still degradation in EIS.

Usually because it requires additional processing and the processors are already at their limits. So, some compromise has to be made. The extent varies depending on how much oversampling is being done.

6K is the minimum to get an approximation of true 4K, but you will need more than that if you have EIS as well. Anything less than 6K raw will get you image degradation after debeyering. An 8K image would allow you to get a close to true 4K final image and have enough headroom to allow EIS. But, as I said, you need the processor headroom to account for all of the extra processing that would be needed.

[Snowbro]

Looks like after all the beating canon took, they are going to be releasing a firmware update that might surprise us in some ways. 

Via canonrumors "Getting information on new firmware updates ahead of time has always been a very big challenge. I assume most firmware development has a soft release date because rushing firmware is never in the best interest of customers. We’re told that a major functionality firmware update for the Canon EOS R should arrive sometime in the first quarter of 2019.

What makes the update “major” and what sort of features are going to be added were not disclosed. It was mentioned that user and media feedback is going to play a major role in future updates for the Canon EOS R and other EOS R bodies. The source did say that it’s possible a minor bugfix firmware update will precede the major release. More to come…"

[thebrothersthre3]

Maybe it'll reduce the crop, which I think they've done on previous cameras.