Noise in negative using NLP compared to normal scan?

Just bought Negative Lab Pro plugin and using it with LR Classic.

Normally i scan my negatives and lets the scanner scan it normally without any processing to get a standard colored scan and postprocess it in Photoshop.

I noticed that after NLP conversion there is a lot of noise in the resulting scan. I am using the default settings of NLP. Compared to a normal way of scan there is just more noise visible. Is this normal or am i doing something wrong?
However, the colors are great and way better than doing it in latest Photoshop CC 21.

I am using latest Vuescan with a Nikon Coolscan V. Vuescan is set according to the guide to save a DNG file. Also tried it with NikonScan but that also has noise in the result. A normal scan don’t seem to show so much noise.

Must i still move the autofocus target in Vuescan when scanning raw negatives?..or is that pointless?

Thanks in advance for any advice.

EDIT: some weird thing i noticed direct after a scan… in Windows File Explorer the DNG file properties says its 120Mb (which it should be) but the resolution is about 255 x 170 / 8 bit/ low dpi while in Vuescan it says output: 5570 x 3736 pixels 4000dpi??? All settings in Vuescan are correctly set as in the guide. Maybe this is why the DNG metadata update is necessary in NLP? I have done that every time before conversion though.

EDIT2: Adobe Bridge is showing correct resolution and dpi for a scanned image though.

I usually notice excess noise on images where I didn’t get the exposure correct in camera, where I underexposed the photograph. As soon as I pull up the shadows or overall exposure without NLP in say Photoshop, I notice a ton of noise. However, when I scan in a perfectly exposed photo, there is not much noise at all, if any, beyond the grain of the film.
It’s worth noting that I too am using VueScan through a Nikon CoolScan V ED.
I used to use CF System’s negative inverter plugin for Photoshop, and had way worse performance through that, so I switched back to good ol’ Epson software and got better results before learning about NLP. I’ve only used the trial of NLP so far, but found better results than both CF Systems and Epson - I re-scanned film that I initially scanned with one of those other programs, so that I could make a direct comparison, and NLP is definitely better at dealing with the scans. Whatever magic it’s doing behind the scenes renders shadows far better, so in my 12 scans I found that unless a shot was incredibly underexposed, I didn’t have to pull shadows up as much, therefore I’m not dealing with noise issues as much.
Now, one of the things worth checking in Vuescan is the “Colour” tab, making sure that the black points, white points, curve high, curve low, and all the brightness sliders are all where they should be. I keep them set to default, but maybe someone else can recommend better settings. I’ve also gone through and set my colour spaces as follows:
Scanner colour space: default
Printer colour space: Adobe RGB
Output colour space: Adobe RGB
Monitor colour space: sRGB (I did notice that the preview changes drastically if you select other options here)
I hope any of this helps!

Question, how are you getting Bridge to show the resolution? Mine will only display the pixel dimensions, but has “- -” where it would normally tell me resolution.

Thanks for the suggestions. It might be called grain or noise what i see. I noticed some other film negatives have less noise, so maybe it also depends on the quality of the strips also?

When i had a Nikon 5000 i had scanned all my negatives already in a normal way without any corrections in NikonScan software. I must say if NLP already existed and i was aware i could just scan it raw, i would have done that, but i wasn’t that experienced.
I don’t want to go over through scanning all again so left most of it as scanned done.
I had bought Color Perfect afterwards, but the interface was just too difficult, so left it and did not use it.
I noticed in Photoshop if i use saturation first and afterwards colorbalance the image just looks as good as with NLP; before i did colorbalance and then saturation/vibrance.

In Vuescan i have set all settings according to Nate’s advice. The Color tab was not mentioned in his guide i think, but there i just set the Color Balance to None and Black + White points all to 0 to prevent clipping. As you mentioned, all color spaces are set to AdobeRGB, even monitor, to get the most color gamut. I am scanning at 4000dpi. Also i have turned on Color Pixels in the Color tab to see if there is clipping. Other settings are default on the Color tab. I think all is set correctly without unnecessary processing.

The dpi resolution i think i got it, because i exported the DNG as TIF from Lightroom with 4000dpi and then used Bridge. Windows File Explorer properties on the DNG shows incorrectly 188dpi. You are correct that Bridge won’t show the resolution for the DNG file.

I exported to TIF to fix the scratches and colorcasts in Photoshop. I would like to do it in Lightroom since its non-desctructive but i know Photoshop better than Lightroom.
After the fixes i might continue with the resulting TIF in Lightroom again for any final corrections.

Grain/Noise i will use Neat Image to get rid of most of it. Still a long way to go through all the scanned images though…it takes a lot of time for me to decide which color result is good enough.

If anyone can recommend the best workflow that would be helpful. Thx.

EDIT: i found a great way to get colors fixed on my normal scans using the Curves and configure the Auto to find dark/white points and gray point. Darn…if i knew this before i did not had to use threshold and color sampler through more than 5000 scans…ouch!