Which picture ratio?

[AyeYo]

Actually that's not correct

Try this little experiment...set your P4P up on a table or something and take two photos - one at 3:2 and one at 16:9 and compare them - the 16:9 has a slightly wider field of view - I've just tried this to confirm my thoughts (above)

Not just wider, but TALLER as well.

 

[AyeYo]

There is something fishy going on here and the raw file we are viewing in viewers like Adobe Camera Raw is not actually a raw file at all - the fact that the camera's software can perform this correction real time at 60fps is really impressive though

I think this is the root of the issue. While the actual RAW might fit DJI's stated resolutions, the corrected DNG actually results in more viewable area on 16:9 than in 3:2 - and DJI seems to have been smart enough to reflect this in the GO4 FPV screen, because seeing the width difference (and that 16:9 is indeed wider, and for me also taller) is as simple as setting the drone on the ground and switching ratios while viewing the video feed.

 

[Meta4]

This confirms what I've seen as well. While I agree with Meta4 that DJI's quoted resolutions would suggest 3:2 is 16:9 with more pixels top and bottom, that isn't what actual shooting bears out. In actual shooting, 3:2 is significantly narrower width-wise and slightly shorter height-wise than a 16:9 picture... and this is absolutely baffling.

What does post #17 show?
I shot two frames with the Phantom on a table and aimed at the same target.
It appears to my eyes to be exactly the same width.
Pixel sizes are actual and correspond to DJI's specifications.

 

[andy_k]

What does post #17 show?
I shot two frames with the Phantom on a table and aimed at the same target.
It appears to my eyes to be exactly the same width.
Pixel sizes are actual and correspond to DJI's specifications.

By the same token what do my posts show? I've posted screen grabs of raws which show a marked difference - just out of interest were your shots the jpegs straight off the SD card? That might explain that even more processing has taken place in camera.

I found very early on that pushing the raws in Photoshop caused graining and noisy shadows - using UF raw is a slower process but produces better shadow detail.

 

[AyeYo]

Don't know because the images are either broken or blocked by work computer. I can see everyone else's photos though.

 

[Meta4]

By the same token what do my posts show? I've posted screen grabs of raws which show a marked difference - just out of interest were your shots the jpegs straight off the SD card? That might explain that even more processing has taken place in camera.

I found very early on that pushing the raws in Photoshop caused graining and noisy shadows - using UF raw is a slower process but produces better shadow detail.

Yes... mine are jpg files straight out of the camera.
What are the pixel sizes of the images you are looking at?

 

[AyeYo]

Maybe we need to get 3:2, 4:3, 16:9 jpeg and DNG series, then compare actual final resolutions to see what we're working with.

 

[andy_k]

All the pixel sizes are the same - see post 10 where I overlayed the two images

What appears to be happening is that the in camera software is correcting the barrel distortion (which we normally see with cameras this wide) then cropping the image and rescaling to the correct size

 

[andy_k]

Maybe we need to get 3:2, 4:3, 16:9 jpeg and DNG series, then compare actual final resolutions to see what we're working with.

if you read back you'll see I've posted all of that

 

[andy_k]

what is happening is probably best shown up in this pic....

The drone takes the background pic (same one as used in my earlier posts) and then applies the correction (I've used the Adobe Warp Transform for illustration) and then crops the resulting shot to exclude the area outside the red lines. The resulting image is then scaled back to what we would expect from the 20mp sensor before being written to the SD card.

This isn't very 'honest' as we are not (as advertised) taking a 20mp image - the data being made available to us is roughly equivalent to a 15-17mp image which is being scaled up to 20mp

Adobe Camera Raw doesn't 'see' the whole image, just the area that DJI has told it to view with whatever profile it's embedded whereas other raw viewers do

Try downloading UFRaw and having a look at the difference on some of your own files.

UFRaw - Home

 

[Jesse_M]

I have to crop all my shots in LR for the 1:1 for Instagram. I shoot RAW. I have read and tried to understand all of your posts but to clarify if I want a taller image (since I have to crop for each post) should I try switching to 3:2 to get higher MP pics? Obviously for my pics wide is not really needed which it sounds like 16:9 is.

 

[AyeYo]

So, the conclusion is that 16:9 does, in fact, produce larger final photos?

 

[AyeYo]

I have to crop all my shots in LR for the 1:1 for Instagram. I shoot RAW. I have read and tried to understand all of your posts but to clarify if I want a taller image (since I have to crop for each post) should I try switching to 3:2 to get higher MP pics? Obviously for my pics wide is not really needed which it sounds like 16:9 is.

If you're cropping to square then there's no need to shoot in 16:9. Might as well have more vertical area in 4:3 or 3:2.

 

[Jesse_M]

If you're cropping to square then there's no need to shoot in 16:9. Might as well have more vertical area in 4:3 or 3:2.

Ok, thank you for clarifying. I will try that. Sorry for newb question but which is better of those 2 4:3 or 3:2 for the most MP's?

 

[AyeYo]

Ok, thank you for clarifying. I will try that. Sorry for newb question but which is better of those 2 4:3 or 3:2 for the most MP's?

Well, that's the golden question this thread this about. It SHOULD be 3:2, but doesn't seem to be the case. Only way to tell is to shoot an image of each with your typical settings and then see which one yields more viewable area and larger file size.

 

[Jesse_M]

Good 'nuff!

Thanks again.

 

[andy_k]

Well, that's the golden question this thread this about. It SHOULD be 3:2, but doesn't seem to be the case. Only way to tell is to shoot an image of each with your typical settings and then see which one yields more viewable area and larger file size.

3:2 yields the largest size image as explained by Meta4 in the first post it gives the appearance of a 20mp image
4:3 yields a slightly smaller image
16:9 under some (but not all) yields a slightly wider image that is exactly the same with as the 3:2 image

None of the above images use the full width of the sensor so the claimed megapixel count is very misleading

 

[KevMo Photog]

Everybody! What Andy K said is straight up! If you are not shooting J-pegs you are not seeing the complete area that sensor is capturing! DJI has set it that way on purpose. But I would still rather tweak a RAW file instead of a jpeg if I'm looking for best quality.

 

[andy_k]

Everybody! What Andy K said is straight up! If you are not shooting J-pegs you are not seeing the complete area that sensor is capturing! DJI has set it that way on purpose. But I would still rather tweak a RAW file instead of a jpeg if I'm looking for best quality.

The problem I am seeing with the DNG files (fake raw ) is that they can't be pushed as far as I would normally expect. This is almost definitely due to then having been processed/enlarged by the time we get our hands on them

 

[Jesse_M]

So the DNG's we receive on the SD card are not true RAW photos?