which market is reality capture? Because it not VFX!

Hey,

Happy new year ( right now its not for me!)

First reason i would like to Clarify is my investment in RC was a recommendation from a work colleague at Pinewood Which he was Clearly Wrong!

I really don’t give a F!about the $200 buying the software because that not really the issues here!

It the time i invested on to it that was a complete waist of my time and now 4 weeks or less of work is down the toilet because iv decided to go back to Photoscan and here is why!

My problem wasnt the results!
Iv benched marked the RC vs Photoscan and it’s about 40 to 60% and can do more then photo scan. The core of the Software is good. So you can pad your self on the back for that one, plus and minus some miss elements on eyes which for some odd reason im willing to forgive! Even though most other won’t! But what the hell!

Every after that is hell on earth.

UI is hordens! and for some od reason look like Microsoft word! And can give you RSI Especially if you need to use it for 8 hours a day for 3 week on close to 100 set of photogrammetry and each of them 170 photos! Selecting anything in the UI makes me want to shove my wacom pen into my eye! Which i still need my eyes if i want to read read the teeny tiny Fing Font which now we use windows magnifying app to read the Fing Parameters! The fact that the view port option is on a tab that i can only access my leveling another tab is fing Retarted! I can go on but i feel like it will take 10 pages!

But that was forgivable because i like the results until it came to the point of dealing with the the geo, filtering and texturing and import and export after is what i can’t forgive.

No EXR support or Image Swaps!: well it clearly the my use is for VFX so i have specific color pipeline for the texture and color workflow that i would like to use! But nope! No exr support! So i end up using Tiff which are to slow and to big! Insted we opted with using jpeg to solve are images and export to maya and nuke to build are textures!

But this isn’t going to work because i Can’t Export your lens distortion to a stmap! So on top of this i need to build a translator for the XMP files just to be able to undistort in nuke to builded the text but nope!

So here the final solution!

Export everything to maya! Including the undistorted jpg
Setup and and export to nuke.
Build a difference map Between the raw and jpg undistorted to apply the distortion to the raw data with all the color information needed!
Connected all of the images to cameras! ( 170 of them!)
Project!

But this would work!
If you’re wondering why! IT BECAUSE THE EACH IMAGE HAS A F@#$%^#$^%#$ OFF SET TO LINE THEM UP! SPEND 4 HOURS TRYING TO FIGURE OUT WHY! AND I FOUND THE F@#@# OFF SET IN THE IMAGE PLANE! AAAAAAAAAAAAAAAAAAAAAAAAAAA

So what the next solution?

We’ll import the retyped geo in to rc and Texture it that way? And that not going to work because!

RC NEED SOME F@#%@#$%#$ .rcInfo TO PLACE THE FING MODEL INTO THE CORRECT POSITION! WHY THE F@#$@#$ ARE YOU NOT JUST USING NORMAL WORLD SPACE POSITION! IT IMPROVES IT WRONG SOO MY F@#$@#$@ TIME WAS WASTED USING THIS PEAC BS %!@#%!#$%!#$%^!@#$^$%^^#!@$@#%@$^@#$%^@#$%!

This software was never ready for prime time!

Im done!

Keep my money i dont Give a F!!!

3 weeks work and it’s unusable!

BYE!

Before blame developer, will be better if you wrote your target.
Because looks like you want simulate Photoscan workflow in more modern software (that just not required 75% of this steps). Or your case is 0.0000000001% of all possible workflow on world.

About UI/UX. You probably never seen how newbie hate zBrush for its UX. But later love it. It not similar to most of apps, but have its own good idea, that help artist work fast and easy.

My work flow is basically most of SOHO VFX, houses work flow. show me how you would work with a software that dosent imposrt the same world position as Maya or max or modo and dosnt support UDIMS!

I pay my artist 25GBP a hour! so im burring money for them to click around! rely! and have to pay RND to build ■■■■ that sould had been there! come on!

15kx25 licensee for a Software for a pro house (which is what the big boys have to pay) no way!

and iv got to test and designing this workflow for a big house! specially on a show where you have 15 Digi doubles!

You pay so much money, but looks like completely ignoring some important steps.

Proper Photogrammetry/3Dscan image acquiring: CameraRigs/Delightening balls/XRite targets/Proper way for shoot images. Preprocessing… etc.

Color matching pipeline: Nuke/Lightroom/Photoshop/3DLut Creator Pro/etc
Scan cleanup and retopology: zBrush/Mudbox/3DCoat/Warp 3

All this steps is important, and big studios use all of them.

dude what are you talking about!
its almost as if we are having a different conversation!
all are rig and setup is fine!
we got every thing we need to do are job! you missed the point! getting in and out of the software is hell! you cant import models because the software uses its won world position!

did you read the post? or are you just replying for the sake of it!

Vladlen wrote:

You pay so much money, but looks like completely ignoring some important steps.

Proper Photogrammetry/3Dscan image acquiring: CameraRigs/Delightening balls/XRite targets/Proper way for shoot images. Preprocessing… etc.

Color matching pipeline: Nuke/Lightroom/Photoshop/3DLut Creator Pro/etc
Scan cleanup and retopology: zBrush/Modo/3DCoat/Warp 3

All this steps is important, and big studios use all of them.

Hi vfxronin

  1. Your style of writing says a lot about you… ok, at least we have an idea…

No EXR support or Image Swaps!: well it clearly the my use is for VFX so i have specific color pipeline for the texture and color workflow that i would like to use! But nope! No exr support! So i end up using Tiff which are to slow and to big! Insted we opted with using jpeg to solve are images and export to maya and nuke to build are textures!

But this isn’t going to work because i Can’t Export your lens distortion to a stmap! So on top of this i need to build a translator for the XMP files just to be able to undistort in nuke to builded the text but nope!

Images can be swapped just by moving the existing images and on the next opening of the project RC will ask you where are the files.
Or just use a good TXT editor and change it in *.RCPROJ file and voila it is done…

But this isn’t going to work because i Can’t Export your lens distortion to a stmap! So on top of this i need to build a translator for the XMP files just to be able to undistort in nuke to builded the text but nope!

Not sure what is STMAP, can you clarify it for us ?
exporting and undistorting images over FBX is quite easy, it is just about inspecting all options :smiley:

FBX_cam_export.png

or even the direct export to Maya with proper setting of image planes and so on

Maya_export.png

But this would work!
If you’re wondering why! IT BECAUSE THE EACH IMAGE HAS A F@#$%^#$^%#$ OFF SET TO LINE THEM UP! SPEND 4 HOURS TRYING TO FIGURE OUT WHY! AND I FOUND THE F@#@# OFF SET IN THE IMAGE PLANE! AAAAAAAAAAAAAAAAAAAAAAAAAAA

So what the next solution?

We’ll import the retyped geo in to rc and Texture it that way? And that not going to work because!

RC NEED SOME F@#%@#$%#$ .rcInfo TO PLACE THE FING MODEL INTO THE CORRECT POSITION! WHY THE F@#$@#$ ARE YOU NOT JUST USING NORMAL WORLD SPACE POSITION! IT IMPROVES IT WRONG SOO MY F@#$@#$@ TIME WAS WASTED USING THIS PEAC BS %!@#%!#$%!#$%^!@#$^$%^^#!@$@#%@$^@#$%^@#$%!

Can you please clarify it for us as a professional and not like an angry child ?
Plenty of users do not need to do anything with the GEO to import it into proper position, again, it is about proper workflows…
And like I say, if you are a true professional, you will start asking proper questions if you are not sure how to solve them.

P.S. then how is it possible that big VFX houses use RC without any problems ? could it be because they ask questions how to solve particular workflows and they are not whining ? answer it to yourself…

Hi vfxronin
reading you last comment. please calm down, take a coffee or smoke a cigarette and come back when you can think like a professional again.

Well. You probably first who have so bad feelings about RC. I know big studios that successfully migrate from obsolete software to RealityCapture. Studios that work in AAA projects.

Just as any software it can required adjusting your current workflow.

And another point. RealityCapture still in development. And evaluate every month. So some required features can be added soon.

Also i think as Pro client you can discuss with developers about important but missed things. May be possible to find any workaround or you just missed some points in software manual. Them always will help you.

Wishgranter wrote:

Hi vfxronin
reading you last comment. please calm down, take a coffee or smoke a cigarette and come back when you can think like a professional again.

hey, Ill man up take back 90% of what i said yesterday and apologises for my Beahiver.

i would still say UI is one issue which broke the camels back and its very hard to access the help dock and software at the same time would be nice to load the help doc in to chrome so i can keep it in a other floating windows to read, and the text sizes are also hard to read because the text are to small specially if your dyslexic ( that should answer to your query to my writing style).
that dose make it hard to work with the software.

I will say result are phenomenal and cant argue with good code.

Hi.

So the biggest drawback I find with camera exports is that Reality Capture does not allow input of a Film back. Regardless of resolution, you should not need any type of imageplane offset if the input image represents the filmback size.

Most applications do not correctly support imageplane offsets so it’s very frustrating when you want to take the RC camera data into Maya and use it in Mari or Mudbox.

The Maya export has come a long way, but I do believe it still needs a lot of work.

Ideally this should be part of the un-distorted images that you export RC.

The offset is required as the principle point of each lens is different, also in reconstruction this is mostly optimized to get a better projection of paired points from other views.