R5 Time-Lapse Gear Updates - What's Up with YouTube?

Mike Lewis

Staff Member
So I was out yesterday late in the day and shot a couple more test time-lapses I will call them. Technically they came out well I think but the subject isn't compelling. I was planning on posting them here anyway, but even after letting YouTube 'work' on them all night, the 4K one is still not rendered in even HD yet, and the HD one that says it is completed looks like fertilizer. So more about that in a minute.

Gear-wise, I tried out my power bank and 1 TB SD card. They both performed well. I was out for about 1.5 to 2 hours, and it was chilly bordering on cold. I had a fully charged battery in the camera at the start and it now shows the same (98%). I shot a 25 minute 4K video time-lapse ~300 images) and then did a matching 300 image time interval sequence (to use Ben's terminology). I also started a sequence running the camera over WiFi from inside my car using the ControlMyCamera phone app (qDSLRDashboard on an Android-based phone) and the intervalometer it supplies, but the interval timing was not accurate with that one so I stopped after about 80 frames. So not a huge amount of frames taken - some number over 700 or so, but the camera battery was still fresh and the power bank did not show hardly any usage either. I am not sure how much the WiFi gets into the battery on the R5 but I am guessing it is normally noticeable. So the power bank with the R5 is a huge success. I hung it from the hook underneath the tripod in a spare lens case.

The 1TB card is still mostly empty too having used 17.4 GB out of an available 954 GB after in camera formatting and the like. The available images was still showing 9999 at the end of the session. The card speed is such that it seems to work fine with a 3 second interval, but it may not be quite fast enough to reliably support a 2 second interval without dropping frames. Of course, since I do not (as of yet anyway) have an external intervalometer to use it is impossible to set anything shorter than a 2 second interval in any case. Using this card the camera will not allow an 8K video timelapse to even be started (requires a CFExpress card), but I do not have any aspirations of making one of those at this point, since 8K videos are almost impossible to process and display properly from what I have read. So with this card and the power bank I would appear to be realistically unconstrained on how long I shoot time-lapses, which is what I was hoping for. Now to just go find a compelling subject.

As far as YouTube, I have lots of questions...

1) Is it normal for this stuff to take so long to convert, or is there something lwrong with my 4K video? It looks and plays fine locally on my 4K monitor.

2) The 4K video ended up at about 500MB for a 10 second clip with no audio. It was created in camera of course. I am assuming that is normal size for a duration like this? (~10 seconds at 30 FPS).

3) Not only do the latest videos I uploaded look like crap when played back on YouTube, but also now the ones I uploaded recently from Sprague Lake. @Ben Egbert - seems like you might have been mentioning something like this as well previously? The quality is so poor I do not plan to even post these unless either YouTube eventually renders them at a decent quality or I can find another place to host them.

4) Anyone have alternatives to YouTube? I am not impressed so far. Maybe a paid account would eliminate the problems? It is too much work to make these to then see them looking like someone smeared Vaseline on the camera lens before you started capturing your images.

ML
 

Jameel Hyder

Moderator
Staff member
The info about external battery is very promising. I might grab one, not specifically for time lapse but as a general backup. The LP-E6EN are pretty expensive and for the cost of a couple, this bank can provide power equivalent to a handful.

Try vimeo for video uploads, it is better than youtube, IMO.
 

Ben Egbert

Forum Helper
Staff member
Hi Mike, thanks for the detailed review, it's great to have all the information we can get on this subject. I am glad your power bank is working so well, I may need to take another look at that.

As to Youtube, my experience is the same as yours, sometimes 2 days to get to 4k, and usually an hour to just get a std def upload.

The terminology I use is the same as Canon uses in the menus and manual.

You have confirmed the 2 second limit for time lapse videos but you can do one second intervals in time intervals.

I agree, 8K is a waste of time at present. I shoot everything in 4K. The set up frames will be jpg and are very good quality.

I set picture style to Landscape with a bit of added clarity and sharpness. I suspect over time I will refine this.

PS can you post another link to power bank? I lost track of the old one and google turns up several of various prices.

@Mike Lewis
 
Last edited:

JimFox

Moderator
Staff member
Hey Mike,

Where is the timelapse? :)

Share the links, trust me Ben and I have shared plenty of videos as tests that didn't have the best quality. At least yours are good quality, you just have to wait for YouTube to render it. We did find that if you change the playback down to 720p once, it will load the HD version and then you can switch back to 4k and it will play fine from there.

Yes, it certainly takes YouTube a while for the HD version to load, I have only noticed it loading it loading first an SD version, then an HD version. I hadn't noticed that it loaded a 3rd version, 4k. Maybe I missed it? Most of mine I think my HD version is loaded within 3 to 4 hours. You just have to be patient and let it do it's thing.

My playback from YouTube will sometimes have an issue rendering the first 11 secs as very blocky, like old 8 bit graphics. But after 11 secs the HD version starts. It doesn't always do it. I have never seen it render a complete video as blocky, well I guess it could if your video was only 11 seconds long then. But for most of ours, we were doing at a minimum I think about 30 seconds as we were playing around with it. Now, I am trying to shoot enough to get all of my videos in the 1 min to 3 min range.

Again, I haven't seen YouTube render a complete video as blocky, only the first 11 secs. And it doesn't always do that. What I am doing to try and avoid that now is if you look at my timelapses, I place a still image for 10 secs with my titles and such before I transition into the actual timelapse. My hope is that if YouTube is slow to render the HD version it will have the first 10/11 seconds to get it loaded.

Jameel suggested Vimeo, the bad thing with Vimeo is you have to pay for it. There is a free version, but it is with limited upload space, just one of my timelapses can blow throw their quota for the week, so there is no way I could use it even. Also I am using my Timelapses, and Ben is also as a tool to share about FocalWorld by our including links to FocalWorld. So with Vimeo, that would be a waste of time, but with YouTube, people could surf it and come across our FocalWorld video's.

BTW, that sounds awesome about the battery. It's great to see it working so well.

As far as an intervolometer, on the R5 you can't set it to 1 sec intervals? If you want an external one Neewer makes a great one that you can set intervals as short as 1 sec, and it costs around $30.
 

Mike Lewis

Staff Member
Answering my own question:

The YouTube Quality mystery is solved. Somehow, my browser got it in its head that I was a mobile phone as far as YouTube was concerned. So the links were getting remapped to the mobile phone versions of the links, which not only changed the layout of the YouTube pages, but removed some of the controls (like the gear control that allows one to set the resolution.) Once I completely cleared all my browser history and cache and came back up it all worked and looked OK. This was indicated by an 'm.' prefix that was appended to the link automatically.

Now the 4K version of my in camera made via video time-lapse is still not converted to 4K yet. I guess (according to my son) Google (aka YouTube) uses vp9 encoding instead of H.264, which is likely why the files are being converted. I expect vp9 takes less space, but it is not efficient to transcode into.

So anyway, here are 2 videos. The first is the current version of the 4K video, which is available in HD but still converting to 4K. This is from internal video time-lapse function in the R5, 5 second interval, 300 frames, 10 seconds at the 30 FPS provided. I will repost a link to the 4K version should it ever finish conversion. It looks decent in HD but really nice (technically) in 4K. No flicker and nice motion. I am struggling to see why using this in camera approach for time-lapses is not pretty ideal, although I guess there is inherently less dynamic range to play with (being 10 bit video I think) and a lot of the settings are baked in when the video is created in camera.


Here is an HD version from individual frames taken after the first. Scene slightly changed to chase the failing light, same 300 frames, same 5 sec interval. It is HD because the free version of LRTimelapse software I used to put it together and de-flicker it only allows HD and not anything higher rez without buying the paid copy. I must say, when used in concert with Lightroom the process has a number of steps but it is not complicated. It took a little more than 30 mins with me reading how to do it, would be faster once you are used to the steps involved. It also then provides you a very efficient way to edit your video using all of Lightroom's tools, automating the application of edits across all the individual frames by just editing keyframes in Lightroom and then having LRTimelapse interpolate the changes across all the other video frames. I think the final result is comparable in smoothness to the in camera video, with the extra flexibility that I can choose any resolution supported by the still frames of the R5 (8K or less), edit for exposure and everything else in Lightroom, and then export via LRTimelapse to whatever format and FPS that my data supports instead of being constrained by the 3 choices offered in camera by the R5. Since this version is constrained to HD I guess it is a fair comparison to the one above. See what you think about flicker, I think it cleaned it up very well, as LRTimelapse analyzes the frames and calculates very small adjustments that are exported back as Lightroom metadata.


ML
 

Mike Lewis

Staff Member
Looking back at it, YouTube says even the HD version of the first video posted above is not yet completed, although it looks OK in that resolution to me...

ML
 

JimFox

Moderator
Staff member
You got some very nice cloud movement in both Mike.

As to the in camera Timelapse, Ben and I both had come to the conclusion it's so much easier to do it that way. Technically to shoot the individual photos and compiling them later into a timelapse gives you more power since you can edit each individual frame. And it allows you the ability to take a frame to edit as a still if one of the images really captures your eye. But for simplicity, there is just no beating letting the camera create the timelapse itself.

I am doing the majority of mine in camera now. Where I wouldn't is when I am shooting the stars at night to create a timelapse, with 20 to 30 sec exposures you don't end up with near the amount of photos to deal with as when doing it during the day.
 

Ben Egbert

Forum Helper
Staff member
I just ordered the Anker at Amazon.

Mike, those videos are both great and no flicker.

You can bring a video into Photoshop and use ACR filter to edit them. I don't use LR and in fact hate the interface. But if it has antiflicker, I suspect ACR also has it. Now to find it.
 

Mike Lewis

Staff Member
Oh yes, did not mean to imply it was Adobe. My bad. I can see how you might think that from the naming convention. I think that is just to illustrate the tight integration with Lightroom.

Looks like the first posted video has FINALLY rendered in 4K.

ML
 

Ben Egbert

Forum Helper
Staff member
Hey Mike, the thing I dislike about Adobe Lightroom is the file handling in data base style. I never know where anything is. If LRT lest you load your files into its UI from folders of your own creation, I might give it a try.

@Mike Lewis
 

Mike Lewis

Staff Member
Ben,

Interesting. I consider the database part of Lightroom to be one of its best features. It empowers the non-destructive editing at the core of the application which really lets you do any number of edits, backwards and forwards on actual or virtual copies and always be able to get back to any step in the process while never harming a single byte of your raw file. It also lets me quickly find 'all the images I have ever taken in Rocky Mountain National Park' or 'all the images I have taken with my 70-300 L IS lens', or have an instant count of how many images I ever took with my 5DsR,' just to name a few.

As for the folders, I am not sure I understand your issue. Yes, you do have to import images before you work on them. But I always copy images off of my memory card to folders of my own naming convention and organization before I import them, at which point they are externally organized just the way I want as well as being cross referenced and accessible in Lightroom. And if I ever need to move them I just move them and then tell Lightroom where I moved them to and it is happy again.

As for LRTimelapse, it loads in your files from where you specify, but eventually you pass that info on to Lightroom so that Lightroom can import your images and edit then keyframes you have identified in LRTimelapse. LRTimelapse will eventually do some of its own naming and locating of the output time-lapse videos, which it recommends are not located in the same file as your stills you used to make the video. I have not used LRTimelapse enough yet to fully understand how it wants to organize its output files yet.

Don't know if any of that is helpful in any way. Perhaps if I still have not answered your question you can ping me again :)

ML
 

Ben Egbert

Forum Helper
Staff member
Thanks Mike. I have found a tutorial and will see if the workflow looks good to me. I assume that once you finish in LRT, you would have a rendered MPF. At that stage I could do any further editing in PS. I don't have LR and even if I did not have to deal with it's file handling stuff, I am not familiar with the interface.

I work my images from raw to finished and if I don't like it, I start over. Not a big deal to me.
 
Top Bottom