Oh it's sooo much easier than manual aligning now! Thanks a lot for this great video! Can't stand RC workflow/interface sometimes, so chances to stumble upon it myself would be nearly zero =))))
Thanks. The best one yet. Two questions: 1) At around 14:12 setting the ground plane. Is there a way of setting it precisely. For instance, the rails or sleepers will be horizontal with the same height( or z axis). Can I set them as ground control points without knowing precisely the x or y coordinates? 2) Did anyone come up with a shortcut yet, to approve a multiple selection of those green suggestion squares yet?
Sure. It should work with enough clear pictures, good lighting, and being sure you don't change facial expression. It's difficult to do by yourself though. Better to have someone else take the pictures.
That's what you see at the beginning and end. But it wasn't really the point of the video. I have other tutorials on the channel that show the highest quality.
@@3DJapan thanks for the quick reply! I understand it wasn't the point here, just wanted to see the result since you seemed to have an impressive amount of photos and goot matches. Looked at the beginning again : great! :)
Hey Phil, Found out a way to Validate those control points faster. 1) Select the image that has control point 2) Click and Drag the Control Point 3) Press Down arrow to confirm the control points. To cancel, the selection, use right click. It was mentioned here: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-S00_mLfbx6o.html
I have a question. If I follow the new way of making control point, is it also possible unregistered images connected to control point? because if some images were not registered on first hand pre-alignment, I wonder how the control point can match the unregisted images later
I dont get what you get with such alignment as it will only align cameras in current component which are already aligned? We need to align same images that are in both components so the can align with each other. Or am I missing something?
Im wondering about the same... Like you select the component with the more images aligned, and place control points based on the automatically generated alignment cloud. But how the other component with the lesser aligned photos knows about these Control Points you place on the first component? Or this is an amazing new way, but we just think about it in the old way?
Sorry I missed this question for so long. I showed how at 3:03 The buttons at the top left let you choose different layouts, including one that shows multiple images.
I always do it wrong 😂 I make points in both component 1&2 on the same places, and that results in mesh will be duplicated some where weirdly... So only shit loads with points in one component and then allign again
I guess they're all different. RC might do better than other apps with another project. If possible I suggest trying to add some more photos around the "edges", where one component ends and another begins.
does not work for me with differnt pieces and contexte do i forget something? (turntable + lumix s5 in jpeg flat + 50mm at f11 + iso 4000 + sweet daylight exterior & spot light + target + white background + 5 control point + align + normal model ) The components does not merging OR merging but i juste have 2 pieces in one not aligned.
Hello. When using "new way". Let's say I have 3 components. Do I use control points in one, like you did (which has the most images), or I should use it in other components too? Thank you.
I'll start with the one that has the most matches and add the one with the second most, then repeat until there's only one component left. Sometimes if a component only has a couple photos it just won't match to anything and I'll just ignore that one or delete it.
If it helps here's where I asked the same question in their forum. support.capturingreality.com/hc/en-us/community/posts/360042889331-Maximal-Texture-resolution-count-error
Hard to tell without trying myself but there are some sets that just won't merge. Either they need more pictures, more details, or the focus is not good.
@@3DJapan does it matter what camera is being used? I was shooting with Sony A7sii. Maybe the lens i'm using. Anyways I just tried shooting with my iphone with manual settings. It's actually working now haha and it looks great
i have two components having different orientation due to some reason if i merge them the orentation reamins same , anyone know how to change rotation of a single component
That's very interesting, I've never heard of that happening. All I can think is to make sure you have enough pictures on both sides. Maybe if the object is symmetrical the software gets confused? I'd suggest asking on their forum where more people can discuss it. Perhaps show pictures and screenshots. support.capturingreality.com/hc/en-us/community/topics
How can I make the orange lines visible? Sorry but also cant understand the new way. You just selecting control points but you dont make new ones. What is the pointof this? I thought new control points cn register unregistered photos or connect components.
Maybe watch again. I click Create to make a new control point every time. The orange lines aren't really necessary but if you really want them go to the View tab and check the box for Residuals. The point is that it's a lot easier than the old way.
Cars are tricky unless maybe if the car is really dirty or has a lot of rust or something on it. The software needs as much random detail as possible to work. They're also shiny which confuses the software. I did it once by spraying the whole car randomly with dry shampoo, which just wipes off or comes off in the rain.
There are some videos on doing shiny objects. You take 2 sets of photos: one with the normal object for texturing and another after you spray it with some matte substance for meshing when it's not shiny.
@@3DJapan It does Thanks a ton I used a dji Spark 2 set of photos in same flight , obviously also same location One set ,camera straight down Another set 45° , trying to get side view.... I get 2 models, overlapping each other. for a long time I thought alignment was wrong, but later realised that if you only viewed one model , everything was right