Do you have examples?

With me it is most of the time pretty exact as you can see here

What caused me to ask today . . .

I am presently rendering an approx 1 hr 30 min video shot with a go pro Hero 5 set at 1920 x 1080 59.940fps

I am rendering with Sony AVC

Internet HD 1080p 59.940 nvidea NVENC template

The only settings I changed in the template:

Max bps 50,000

Ave bps 28,000

Preset: High Quality

VBR High Quality

My target is less than 20GB. (In an ideal world, VMS would tell me whether and how much to back off to achieve my target file size).

VMS is reporting an estimated file size of 3 MB. I'm expecting much larger

I'm hoping that I have a wrong setting somewhere but I can't find it. But on other occasions VMS has reported a file estimate that was much smaller than the rendered file (but in the past, more like reporting 18GB and the rendered file is 26GB.

Sounds about right for NVENC and internet template.

Of course, estimating hardware encoded file sizes is a crapshoot, best case.

I am rendering with Sony AVC

Internet HD 1080p 59.940 nvidea NVENC template

I'm very curious where you find those settings in Sony AVC, there I don't have NVENC possibilities

I only have those in the Magix AVC codec and I now am trying to render my GOPro 5 files (2,7K) to that option in Magix AVC with your changed settings. Report later.

I may have misspoke about the codec.  I am running a render now so could not double check it.  I am guessing that you are correct, and the codec I am using is found under Magix AVC, not Sony AVC.

I rendered 5 minutes GoPro 5 2,7 K 50p to Magix AVC with NVENC in 1080p 50fps, high quality, max 50 Mbps, aver. 28 Mbps.
Estimated filesize 1008 MB
Rendering for those 5 minutes took 4 min 46 sec and the filesize is 0,98 GB= 1007 MB, looks nothing wrong to me.

My rendered video is 8.96GB (versus the 3MB that was the estimated file size),  So, in other words, VMS15 underestimated the file size by 8.947GB . . . not even remotely close to accurate in my case.  And I experience this phenomenon regularly  Literally, gigabytes of miscalculation.  I get that file size estimating is something of a "crapshoot" as Musicvid as observed, but an estimate that is this far off is beyond the realm of crapshoot.  This is not an isolated occurrence.  It happens perhaps every time I render, but it is only recently that I have been paying attention.

Where are you seeing those figures?

At "make movie" without making a choise or with "advanced option"s where you have to alter the settings?

My rendered video is 8.96GB (versus the 3MB that was the estimated file size), 

I can't imagine if you made a choise for a template that it says 3 MB.

Is it possible to make a screenshot of that 1h 30m timeline and a choosen template like this and which build of VMS 15 are you using??

I was seeing 3GB not MB.

Yes, something is not set correctly.

I've run across this with VMS14. I was only playing around with test videos so didn't take much notice. But I do recall the indicated size was way out to what would be expected. Now using VMS15 I'll take closer notice in future.

There was something wrong with the render . . . Vimeo would not recognize it as a valid codec (frustrating given that I now have to rerender and re-upload a sizeable file).  So I am running it again, but this time I am not changing any settings in the template.  I got a much more realistic size estimate of 13GB so I will see how this attempt comes out.

I've now discovered where I noticed the problem. For me it occurs when using Sony XAVCS export. As per the image the estimated file size is 189mb whereas it's actually over 4,000mb. Interestingly all the XAVCS templates indicate the same estimated size but that could be because they all use the same bitrate irrespective of image size and framerate. Unfortunately the templates don't show any options for bitrate, so you cannot check.

The other area of inaccuracy, with all renders, is when you reach 1gb. VMS only shows the estimated file size to the nearest gb. A 1.4gb file will merely show as a 1gb estimate.

There was some weird glitch that I could never run to ground. In both cases, the template produced mp4 files that were not recognized as videos. But the third time was the charm and produced a video that was smaller than estimated, which I can live with. I am going to keep watching this so I can understand the behavior better.

Curious, but for what purposes do you use the XAVCS templates?

Curious, but for what purposes do you use the XAVCS templates?

I occasionally use it when I need to create an intermediary file of good quality to convert my 50i source material to 50p to get around VMS's poor handling of interlaced material when slowing down clips with resampling disabled. I'd use Handbrake but it doesn't produce proper 50p files.

I'd use Handbrake but it doesn't produce proper 50p files.

Can you elaborate? I don't generally do 50p but I don't recall hearing about an issue with hb.

If I ask Handbrake to produce a 50 fps file from my 50i source I get a file that is 50 fps P. But instead of 50 different frames I get 25 frames duplicated. Handbrake appears to deinterlace but then instead of using both fields to create two frames just uses one field and duplicates it. It is not something that most would do, so unlikely most users would stumble on it. Usually when I use Handbrake I'm converting 50i to 25p for a Vimeo upload.

Eric, converting 50i (25i) to 50p is another filter altogether in Handbrake. The normal frame rate conversion is drop or dupe, period. No blend, no interpolate, no bob.

--> Set your Handbrake output to 50p, constant frame rate. Then use the dropdown menu and select the BOB Deinterlace Filter. You will get georgeously field-interpolated 50p video, with very little added noise.

Historical note: The Bob filter was invented by Donald Graft (neuron2) around 2000, and I first used it when it became available for VirtualDub. A lot of early cards captured 1/2 PAL or 1/2 NTSC SD interlaced back then, so the filter was a huge advantage in those days.

Musicvid you're a marvel. In the Filters tab I set it to Deinterlace: Yadif, Preset: Bob and Interlace Detection: Default. And it worked. I can now batch convert several files should I need to do so. Yippee. Many thanks.

I'm not sure that Deinterlace:yadif has any effect , and I'm glad you discovered this useful tool.

I get a little annoyed with gamers who use it to try to "trick" YouTube into a higher bitrate, however. I couldn't come up with a test that would detect much of a difference, save for a slight, but measurable increase in shadow noise.

I'm not sure that Deinterlace:yadif has any effect , and I'm glad you discovered this useful tool.

It probably doesn't but the options in my 1.0.7 (2017040900) 64 bit version of Handbrake for Deinterlace is Off/Yadif/Decomb. Yadif seemed the most sensible. Off doesn't give me the Preset window to select BOB so that's out. A quick test indicates Decomb gives the same result as Yadif.