Your question is not clear to me.

To calculate the ddCt value of the gene X e.g. untreated samples vs treated sample. Then,

1. You will perform RT-qPCR of untreated sample with endogenous control (e.g Actin/GAPDH) as well as GeneX.Once you get the Ct values, calculate the dCt of GeneX using formula:

dCt (untreated)= Ct (geneX, untreated) -Ct (Actin/GAPDH, untreated)  

2.Simultaneously perform RT-qPCR the same with treated sample, and calculate the dCt for treated sample

dCt (treated)= Ct (geneX, treated) -Ct (Actin/GAPDH, treated)

Here the untreated is a calibrator used to calculate the ddCt of geneX(treated) using the formula:

ddCt(geneX)=dCt (treated)-dCt (untreated)

If you are performing a time course analyses then 0h is usually a calibrator used to calculate the ddCt, This ddCt is then used to calculate the RQ values of the experimental. I have attached ABI's Q-PCR guide for your reference.

Good Luck

If you are using q-PCR analysis to compare between gene expression levels in treated and control groups so the calibrator will be the control.

If you are using q-PCR analysis to estimate expression levels of certain gene in normal condition without any specific treatments then you can use the sample with the highest ΔCt value as your calibrator

I attached a pdf on Real-Time Quantitative PCR Assay Data Analysis, Evaluation and Optimization.

you will find the answer to your question on page no. 18

Good luck

http://www.gene-quantification.de/lanz-qpcr-tutorial-bcm-2009.pdf

yep. whatever gives you the cleanest answer should be chosen. and in some cases, if you have enough replicates and are confident in downstream analysis robustness- you can skip normalization altogether

That the most useful reason for using a calibrator sample is inter-assay variations.

But also expresing a result in terms of fold change (i.e. untrated vs treated or kinetics).