Video from the C23 and H Plus

[PatR]

To retain 20mp resolution after a crop of any size would be a feat of magic. You might trick the EXIF data into stating it was the original size, but the report would be false.

 

[Peggy]

Did I understand @Peggy correctly that Yuneec cropped the JPEG to straighten the horizon but not the DNG???

Thats correct @rdonson DNG have been left totally raw

 

[Peggy]

Here is the colour video again updated, I had to remove the last one due to errors in typos that I made.

 

[Peggy]

Sure, I’ll post some tomorrow as it’s nearly midnight here and I’m in bed now. The best setting if I’m honest is ‘original’ colours come real nice. All the others are oversaturated and a nightmare to work with in post.
Also noticed a big typo error in that video and music is distorted, so will redo and repost tomorrow as well

You asked me to upload some photos from the H+, Here are the ones I used in the colour video completely unedited for you to play with. I would respectfully ask that when any of you have had a play with them you do not share or post on any form of website or social media but simply use for your own testing. Thank you

Photos from H+ for forum

 

[Mickeyboo]

After watching the Pegasus video posted by YuKay it appears there’s some room for a better lens. A pin cushion effect is definitely present.

I think issue may be an improperly loaded profile if they are using Lightroom...I used the Inspire profile for a while but have since turned profile off entirely.

 

[Mickeyboo]

Here is the colour video again updated, I had to remove the last one due to errors in typos that I made.

Thanks for taking the time to do this but overcast flat light is not really good light for shooting this sort of demonstration...the differences are much more dramatic when shooting in sunshine. I always use normal so the DNG has more latitude for dealing with highlights and shadows.

 

[rdonson]

Thats correct @rdonson DNG have been left totally raw

If Yuneec adjusts for the lens characteristics and that results in a crop in the JPG that means that Yuneec has quantified the lens performance. This isn’t the end of the world.

There is no such thing as a perfect lens. Especially one with a 91 degree FOV that is relatively inexpensive.

I have lenses for my Canon and Fuji cameras that cost thousands of dollars that I perform corrections on. Granted the corrections are minor compared to the drone camera lenses but physics is physics.

Adobe even provides the tools so that you can create your own lens corrections for their products if you want to take the time.

 

[PatR]

You asked me to upload some photos from the H+, Here are the ones I used in the colour video completely unedited for you to play with. I would respectfully ask that when any of you have had a play with them you do not share or post on any form of website or social media but simply use for your own testing. Thank you

Photos from H+ for forum

Peggy,

Thanks again for publishing your work for others to view and work with.

You last sentence in the post above was something people should take very seriously and need to honor. I don’t know if people are aware of it but recently copyright law has been extended to cover re-posting of video, photos and other work. People cannot re-post the work of others without permission from the owner and can suffer consequences for doing so.

 

[PatR]

I think issue may be an improperly loaded profile if they are using Lightroom...I used the Inspire profile for a while but have since turned profile off entirely.

Good to know, thanks.

 

[Peggy]

Thanks for taking the time to do this but overcast flat light is not really good light for shooting this sort of demonstration...the differences are much more dramatic when shooting in sunshine. I always use normal so the DNG has more latitude for dealing with highlights and shadows.

You gotta take the shot when you can and have the spare time. Waiting for the perfect sunlight is not a luxury my life has time for. I was available and free to take a flight at that time.
I take your comment on board and will make sure I only post information when the lighting is indeed in my favour.

 

[rdonson]

To retain 20mp resolution after a crop of any size would be a feat of magic. You might trick the EXIF data into stating it was the original size, but the report would be false.

Since Photoshop CS5 and Lightroom you would simply use "Content Aware Fill" to achieve that. Adobe has the magic.

 

[Rubik]

To retain 20mp resolution after a crop of any size would be a feat of magic. You might trick the EXIF data into stating it was the original size, but the report would be false.

I was thinking that perhaps the sensor has more than 20 Mpixels available and the .dng image used more pixels than the .jpg image thus allowing for some cropping to correct for barrel distortion while still producing a 20 Mp final image. While the 1" Sony sensor does have slightly more than 20 million effective pixels (at 3:2 image ratio), that's not what happens. Looking at the data from the Pegasus files, I see that the .jpg files actually have slightly more pixels than the .dng files. Of course, the smaller .jpg file size is a result of the .jpg compression (on the order of 59% for these files).

So, I agree with you. Any cropping, whether in the camera or in post, will reduce the resolution. That's one good reason to have lots of pixels to start with. Perhaps Yuneec or Peau or pixAero will come up with a lens for the C23 that has less distortion. If not, it's still an impressive aerial camera system.

 

[rdonson]

@Rubik as you say, its still an impressive aerial camera system especially for this price point.

I haven't seen anything that would keep me away from the C23 because it can't be solved in post for video or stills.

 

[Rubik]

You asked me to upload some photos from the H+, Here are the ones I used in the colour video completely unedited for you to play with. I would respectfully ask that when any of you have had a play with them you do not share or post on any form of website or social media but simply use for your own testing. Thank you

Photos from H+ for forum

Thanks, Peggy, for posting these photos .
I fiddled with the .dng files and found myself adding more contrast and much less sharpening than the .jpg files from the camera. Also, it was easy to correct the fisheye distortion in the .dng files. As you requested, I'll not post them here or anywhere.
Happy flying!

 

[PatR]

Rubik,

If that sensor had any more to offer than 20mp you can bet your butt Yuneec would have claimed it in their advertising. Take the Anafi as an example, they are advertising a 21mp sensor. The numbers are a big deal in advertisingz

 

[CianMcLiam]

DJI Phantom 4 Pro uses the same kind of digital manipulation in their DNG files but have been much more devious in how it's done. The DNG files when opened in Lightroom look great and are reported as 20MP, however if you open the DNG in RAWTherapee, which does not read the embedded profile, you can see the original captured image has heavy barrel distortion like the C23 has. This is not seen in the JPG or DNG files opened in any software that can read the embedded profile corrections.

Basically Yuneec and DJI crop and correct the image to about 17MP or thereabouts and then up-sample it back to 20MP. Yuneec only seem to apply this to the JPG, DJI do it for both.

 

[Rubik]

DJI Phantom 4 Pro uses the same kind of digital manipulation in their DNG files but have been much more devious in how it's done. The DNG files when opened in Lightroom look great and are reported as 20MP, however if you open the DNG in RAWTherapee, which does not read the embedded profile, you can see the original captured image has heavy barrel distortion like the C23 has. This is not seen in the JPG or DNG files opened in any software that can read the embedded profile corrections.

Basically Yuneec and DJI crop and correct the image to about 17MP or thereabouts and then up-sample it back to 20MP. Yuneec only seem to apply this to the JPG, DJI do it for both.

So, you're saying that the embedded profile directs the viewing software to upscale the image? Viewing the file without reading the embedded profile reveals that the image is actually less than the advertised 20 Mp. Interesting... I'm remembering what Pat said earlier about "lying".
I'm not sure any of this really matters as long as we get the image or video that we want and it stands up to critical viewing on good equipment.

 

[rdonson]

So, you're saying that the embedded profile directs the viewing software to upscale the image? Viewing the file without reading the embedded profile reveals that the image is actually less than the advertised 20 Mp. Interesting... I'm remembering what Pat said earlier about "lying".
I'm not sure any of this really matters as long as we get the image or video that we want and it stands up to critical viewing on good equipment.

This has been normal for a long time in digital cameras. Either Adobe or the manufacturers provide info in the EXIF data on how to make a number of corrections for the lenses. In Adobe Lightroom, for example, you can choose the lens profile manually if you'd like or apply it on import. Most of today's processing software for stills and videos can do this. There are NO perfect lenses even ones that cost thousands of dollars.

 

[CianMcLiam]

So, you're saying that the embedded profile directs the viewing software to upscale the image? Viewing the file without reading the embedded profile reveals that the image is actually less than the advertised 20 Mp. Interesting... I'm remembering what Pat said earlier about "lying".
I'm not sure any of this really matters as long as we get the image or video that we want and it stands up to critical viewing on good equipment.

Yes, the raw file is only like 1010101010101001010111001 etc., the camera manufacturer can determine quite a bit about how software renders the final image. Cropping and interpolation seems to be relatively easy for manufacturers to apply. In fact, I think even standard raw files are subject to colour interpolation since each pixel is just one colour.

 

[PatR]

So, you're saying that the embedded profile directs the viewing software to upscale the image? Viewing the file without reading the embedded profile reveals that the image is actually less than the advertised 20 Mp. Interesting... I'm remembering what Pat said earlier about "lying".
I'm not sure any of this really matters as long as we get the image or video that we want and it stands up to critical viewing on good equipment.

That's why I went with a 920 and the 16mp CGO-4. No Tomfoolery in having crop for image corrections. I have yet to see any improvement in imagery with the E-90 over the CGO-4.

As for "lying", we might want to take a step back from that word. I'm sure the sensor records at 20mp, which is what they state it does. What happens after that I don't recall them talking about.